Phylogenetic and structural insights into the origin of C-type lectin Mincle in vertebrates

Mincle homologues in lower vertebrates

To find proteins homologous to Mincle in non-mammalian vertebrates, we searched for molecules with similar characteristics to mammalian Mincle in terms of amino acid sequence. Mincle in mammals shares several characteristics in their primary structures, namely the EPN (Glu-Pro-Asn) motif for recognition of glucose and mannose and a hydrophobic groove that binds acyl chains of glycolipids (Feinberg et al. 2013); (Furukawa et al. 2013).

We searched for human Mincle (huMincle) homologues in fish, one of the evolutionarily farthest vertebrates from mammals. In Takifugu rubripes, one of the first vertebrates whose whole genome was completely sequenced, a protein sequence bearing the EPN motif and a hydrophobic groove was found (XP_029706534.1) (Fig. 1A). Other bony fishes also had protein sequences that bear these Mincle characteristics, like T. rubripes (e.g., C-type lectin domain family 4 member E-like in Polyodon spathula (Mississippi paddlefish) and hypothetical protein SKAU_G00319680 from Synaphobranchus kaupii (Kaup’s arrowtooth eel)) (Fig. 1B). We further BLAST searched for Mincle homologues in sequences across vertebrates except for mammals. In the top 1000 hits, protein sequences from reptiles, amphibians, and bony fishes were included (Supplementary Table 2) and some possessed the amino acid sequence characteristics of mammalian Mincle. These results suggest that Mincle homologues possessing an EPN motif and hydrophobic groove might be present throughout jawed vertebrates.

Fig. 1figure 1

Mincle homologue in Takifugu rubripes. A Amino acid sequence alignment of full-length Mincle and FcRγ from human and T. rubripes. The transmembrane region and the CRD are shaded in gray and yellow, respectively. Amino acid residues that interact with a calcium ion are shown in red, hydrophobic groove-forming residues are shown in blue, and the characteristics of the mammalian Mincle are indicated in bold. The immunoreceptor tyrosine-based activation motif (ITAM) is labeled below the amino acid sequence. B Amino acid sequence alignment of full-length huMincle and Mincle homologues in P. spathula and S. kaupii. The transmembrane region and the CRD are shaded in gray and yellow, respectively. Amino acid residues that interact with a calcium ion are shown in red, hydrophobic groove-forming residues are shown in blue, and residues that are defined as the characteristics of the mammalian Mincle are indicated in bold. C Flow cytometry analysis showing the cell surface expression levels of HA-tagged Mincle and FLAG-tagged FcRγ on HEK293 transfectants. The species are indicated above. Data are representative of three independent experiments

As mammalian Mincle relies on FcRγ to signal, we further searched for FcRγ in T. rubripes. Indeed, there was a FcRγ homologue containing a conserved ITAM motif, consistent with the previous study (Guselnikov et al. 2003) (Fig. 1A). We then BLAST searched for human FcRγ (huFcRγ) in lower vertebrates, which revealed its sequence to be also well-conserved in reptiles, amphibian, and fishes, inferring the presence of FcRγ interacting CLRs, like Mincle (Supplementary Table 3). This suggests that in evolutionarily distant non-mammalian vertebrates, specifically fish, Mincle could interact with FcRγ to function, similar to mammalian Mincle.

As we found Mincle and FcRγ-like protein sequences in T. rubripes, we next set out to confirm the interaction of these proteins in vitro. Mammalian Mincle is not expressed on the cell surface without the interaction with FcRγ (Yamasaki et al. 2008). Thus, we ectopically expressed T. rubripes Mincle (trMincle) in the presence or absence of T. rubripes FcRγ (trFcRγ) in HEK293 cells and compared the surface expression of trMincle. Cell surface expression of trMincle was detected only in the presence of trFcRγ (Fig. 1C) suggesting that trMincle and trFcRγ behave similarly to mammalian Mincle and FcRγ in terms of surface expression at least in mammalian cells.

Altogether, trMincle shares similar characteristics with mammalian Mincle in its amino acid sequence and its FcRγ-dependent cell surface expression. This is despite the fact that trMincle is automatically annotated as a CD209 antigen-like protein A by the gene prediction method, Gnomon. However, the characteristics such as the short intracellular region without any signaling motifs, the conserved hydrophobic groove-forming amino acids, and the FcRγ-dependent cell surface expression of trMincle suggest that this protein is more likely a homologue of Mincle rather than CD209.

Structural analysis of trMincle

Next, to compare the 3D structural similarity of trMincle with mammalian Mincle, a structural analysis of trMincle was performed. As trMincle was predicted to be a type II transmembrane protein (Supplementary Fig. 1A, B), we constructed the extracellular domain of trMincle as a soluble recombinant protein (Fig. 2A) followed by column-based purification. The crystal structure of trMincle was determined at 1.7 Å resolution (Fig. 2B). A calcium ion was coordinated with Glu-Pro-Asn (EPN) and Trp-Asn-Asp (WND), two well-conserved motifs, suggesting the typical sugar-binding pocket observed in other mammalian C-type lectins was also present in trMincle. Furthermore, one glycerol molecule in the media interacted with the EPN motif within the hydrophilic pocket (Fig. 2B), and hydroxyl groups at C1 and 2 of glycerol formed coordination bonds with calcium ion (Fig. 2C), which is similar to the reported sugar-binding mode of mammalian Mincle (Feinberg et al. 2013).

Fig. 2figure 2

trMincle is structurally similar to boMincle. A Schematic representation of full-length trMincle (upper) and soluble trMincle (lower). TM, transmembrane helix; CRD, carbohydrate recognition domain. B Overall structure of the trMincle CRD homodimer in complex with glycerol (PDB ID: 9KS7). Protein, glycerol, and calcium ions are shown in ribbon, stick, and sphere models, respectively. C Close-up view of the glycerol binding site in trMincle. The hydroxyl groups 1 and 2 of the glycerol are labeled in gray. Coordination bonds are indicated by brown dotted lines, and hydrogen bonds are indicated by black dotted lines. D List of the top ten PDB structures that are structurally similar to trMincle obtained from a full PDB search of Dali server. PDB IDs and chain names are indicated in the chain column. Descriptions of each structure are indicated in the PDB Description column. Mincle in the list is highlighted in red. E Surface representation of overall ligand binding region of boMincle (PDB ID: 4KZV) (left panel) and trMincle (PDB ID: 9KS7) (right panel). Schematic drawings of the receptor showing representative grooves are given above the panel. Asterisks indicate the corresponding grooves in the structure. F Representative state of trMincle ligand binding region inferred from molecular dynamics simulation. The left panel is the initial state, while the right panel shows the final state of the simulation. These images are depicted from the same view angle. G Comparison of the position of the hydrophobic groove on boMincle (left panel) and the putative hydrophobic groove on trMincle in the final state (right panel). Calcium ion is represented as a gray sphere

Using this first structure of Mincle homologue in non-mammalian species, we conversely searched for homologues based on structural similarity. As expected, a Dali server search of all deposited structures in the Protein Data Bank (PDB) revealed that mammalian Mincle, such as the bovine Mincle (boMincle) CRD (PDB ID: 4KZV), was the most structurally similar protein to the trMincle CRD (Fig. 2D). Indeed, overall, both fish and mammalian Mincle possessed hydrophilic pockets and adjacent grooves (Feinberg et al. 2013); (Furukawa et al. 2013), although the amino acid residues were not identical (Figs. 1B and 2E, Supplementary video 1, 2). Moreover, the molecular dynamics simulation suggested that the trMincle CRD can create a typical groove similar to a mammalian Mincle (Fig. 2F, G, Supplementary video 3).

trMincle possesses narrow sugar-binding pocket

Mammalian Mincle binds calcium ion via an EPN motif which is responsible for its sugar-binding ability. As the structural analysis suggested that trMincle possesses a sugar-binding pocket (Fig. 2B), we set out to investigate the sugar recognition mechanism using monosaccharide glucose as a candidate for the model ligand. Indeed, the crystal structure of the trMincle CRD-glucose complex was determined at 1.8 Å resolution (Fig. 3A). As suggested by the glycerol complex, one glucose molecule interacted with the calcium ion as well as the surrounding side chains (Fig. 3B). Thus, the sugar-binding mode of trMincle was similar to the other CLRs bearing EPN motifs (Drickamer and Taylor 2015) (Fig. 3B). In contrast, the sugar-binding pocket of trMincle was narrower than that of mammalian Mincle, such as boMincle (PDB ID: 4ZRW) and huMincle (PDB ID: 3WH2) (Fig. 3C), suggesting that trMincle preferentially binds monosaccharide, not disaccharide. This is in marked contrast to mammalian Mincle, which has the capacity to recognize trehalose disaccharide (Ishikawa et al. 2009); (Feinberg et al. 2013).

Fig. 3figure 3

Structural differences in the sugar-binding pocket. A Overall structure of the trMincle CRD homodimer in complex with glucose (PDB ID: 9KPL). Proteins, glucose, and calcium ions are shown in ribbon, stick, and sphere models, respectively. B Close-up view of the glucose binding site in trMincle. The hydroxyl groups 3 and 4 of the glucose molecule are labeled in gray. Coordination bonds are indicated by brown dotted lines, and hydrogen bonds are indicated by black dotted lines. C Distances of amino acid residues shaping the sugar-binding pocket. The measured distances are indicated near the dotted lines. D Overlayed structure of trMincle and boMincle-trehalose complex. Structures were superposed according to the position of the calcium ions in their sugar-binding pockets. Surface of the boMincle and trMincle is shown in orange and blue, respectively. The calcium ion is shown as a gray sphere

Indeed, computational superposition of the boMincle-trehalose complex onto the trMincle-glucose complex demonstrated that trehalose causes a heavy steric clash with trMincle, suggesting that trehalose disaccharide cannot bind to trMincle due to the narrow pocket (Fig. 3D). It evokes the possibility that trMincle cannot recognize disaccharide-bearing glycolipids, which are typical of exogenous origin, such as mycobacterial glycolipids (e.g., TDM or trehalose monomycolate (TMM)), but instead may only bind monosaccharide-bearing glycolipids, which are often derived from self, like β-GlcCer (Nagata et al. 2017). While mammalian Mincle recognizes both exogenous diglycosylated lipids and endogenous monoglycosylated lipids, our results imply that trMincle may function as a receptor that selectively binds ligands such as monosaccharide-bearing glycolipids.

Comparing the structural models of Mincle homologues across reptiles and amphibians

As mentioned, proteins with characteristics of huMincle are conserved among several species. To gain an insight into how Mincle evolved between fish and mammals, we first searched for homologous proteins in amphibians and reptiles respectively using protein BLAST (Supplementary Table 4, 5). Phylogenetic analysis of the top 10 hits of amphibians and reptiles was conducted to check the amino acid sequence similarity to huMincle and boMincle (Fig. 4A). All reptile-derived molecules and four out of ten amphibian molecules were classified in the same clade with mammalian Mincle. As the other six amphibian homologues were grouped into the same clade as trMincle, structure predictions with AlphaFold3 were performed. Dali server was used to analyze the overall structural similarity between predicted structures and the huMincle and boMincle crystal structures (PDB ID: 3WH2, 4KZV). Dali Z-score, a score that provides information on structural similarity, was higher for boMincle than that of huMincle, in both species (Fig. 4B, C, Supplementary Fig. 2A, B). Amphibian molecules were divided into two clusters based on their Dali Z-score (Fig. 4B). The higher Dali Z-scored proteins were in the same clade as mammalian Mincle, while the other lower Dali Z-scored proteins were in the same clade with trMincle (Fig. 4A, Supplementary Fig. 2A).

Fig. 4figure 4

Evolution of ligand binding region of Mincle CRD. A Maximum likelihood tree for amino acid sequences inferred from the BLAST top 10 hits for searching Mincle homologues in amphibians and reptiles. The numbers beside each node indicate bootstrap values calculated with 1000 replications. Amphibian-derived molecules and reptile-derived molecules are highlighted in cyan and green, respectively. Molecules in the same clade with trMincle are indicated as the trMincle group, and those in the same clade with huMincle and boMincle are indicated as the mammalian Mincle group. Human Dectin-1 (huDecin-1) was used as an outgroup. Dali Z-score distributions of B amphibian Mincle homologues and C reptilian Mincle homologues in Dali search. Dali Z-scores were calculated by comparing the structures of the AlphaFold3-predicted CRD structure of Mincle homologues and the crystal structure of huMincle or boMincle. D Superposition of boMincle (PDB ID: 4KZV) (brown) and amphibian Mincle homologues (cyan). The structures were superposed according to the positions of calcium ions in their putative sugar-binding pockets. Structures in trMincle groups and mammalian Mincle groups are shown in the left and right panels, respectively. Individual structures viewed from the same angle are shown in Supplementary Fig. 2C. Trehalose bound in the sugar-binding pocket of boMincle is shown in a stick model. Calcium ion is shown in a sphere model. E Superposition of boMincle (PDB ID: 4KZV) (brown) and reptilian Mincle homologues (green). The structures are superposed according to the positions of calcium ions in their putative sugar-binding pockets. Structures that have steric clashes with trehalose (clash (+)) and those without steric clash (clash (–)) are shown in the left and right panels, respectively. The arrow in the clash (+) panel indicates the location of the clash observed in the structure. Structures of the sugar-binding pocket in individual reptilian protein viewed from the same angle are shown in Supplementary Fig. 2D. Trehalose and calcium ions are shown in a stick model and in a sphere model, respectively. F Scheme that shows the widening of the sugar-binding pocket and the broadened capacity of expected ligands during evolution

Next, to compare the sugar-binding pocket, we superposed trehalose-bound boMincle CRD structures (PDB ID: 4KZV) onto predicted structures of amphibian and reptile homologues, by aligning the sugar-binding pockets’ calcium ions (Fig. 4D, E). Amphibian Mincle in the same clade as the mammalian Mincle had broad sugar-binding pockets, big enough to bind disaccharides (Fig. 4D, Supplementary Fig. 2C). In contrast, amphibian Mincle in the same clade with trMincle had narrower binding pockets, similar to trMincle (Fig. 4D, Supplementary Fig. 2C). Reptile Mincle had, overall, broader sugar-binding pockets than trMincle; however, some still had steric clashes with a trehalose (Fig. 4E, Supplementary Fig. 2D). In sharp contrast to mammalian Mincle that possessed a wide sugar-binding pocket, most of the fish Mincle were predicted to possess narrower pockets (Supplementary Fig. 2 E, F). Proteins with or without steric clashes were further classified as two distinct groups according to the phylogenetic analysis (Fig. 4A, E). These data suggest the widening of the sugar-binding pocket during evolution.

Comments (0)

No login
gif