Structure of the human NK cell NKR-P1:LLT1 receptor:ligand complex reveals clustering in the immune synapse

Abstract

Signaling by the human C-type lectin-like receptor, natural killer (NK) cell inhibitory receptor NKR-P1, has a critical role in many immune-related diseases and cancer. C-type lectin-like receptors have weak affinities to their ligands; therefore, setting up a comprehensive model of NKR-P1-LLT1 interactions that considers the natural state of the receptor on the cell surface is necessary to understand its functions. Here we report the crystal structures of the NKR-P1 and NKR-P1:LLT1 complexes, which provides evidence that NKR-P1 forms homodimers in an unexpected arrangement to enable LLT1 binding in two modes, bridging two LLT1 molecules. These interaction clusters are suggestive of an inhibitory immune synapse. By observing the formation of these clusters in solution using SEC-SAXS analysis, by dSTORM super-resolution microscopy on the cell surface, and by following their role in receptor signaling with freshly isolated NK cells, we show that only the ligation of both LLT1 binding interfaces leads to effective NKR-P1 inhibitory signaling. In summary, our findings collectively support a model of NKR-P1:LLT1 clustering, which allows the interacting proteins to overcome weak ligand-receptor affinity and to trigger signal transduction upon cellular contact in the immune synapse.

Fig. 1. Sequence alignment of human NKC-encoded receptor:ligand pairs showing shared structural and functional sequence motifs.

Secondary structure elements and loop regions (L) are denoted for NKR-P1 and LLT1 above the alignments. The paired numbers at the bottom indicate the disulfide pairs in the NKR-P1 and LLT1 structures; asterisks mark His176Cys mutation in LLT1 and Ile168 residue in NKR-P1. The predicted N-glycosylation sites of NKR-P1 and LLT1 are denoted with orange triangles. The conserved WIGL motifs are underlined in black. Blue lines above the sequence indicate the regions forming the non-covalent dimers of NKR-P1 or LLT1. Conserved residues are marked red; bold letters denote strictly conserved residues. a Sequence alignment of CTLDs of human NKR-P1-related NK cell receptors, i.e., human NKR-P1, NKp65, and NKp80. NKR-P1 residues contacting LLT1 in the NKR-P1:LLT1 complex in the primary binding mode, and NKp65 residues contacting KACL in the NKp65:KACL complex, are highlighted in green. Purple triangles indicate NKR-P1 residues that engage LLT1 in the NKR-P1:LLT1 complex in the secondary binding mode. b Sequence alignment of CTLDs of LLT1-related human CLEC2 ligands, i.e., LLT1, KACL, and AICL. LLT1 residues contacting NKR-P1 in the NKR-P1:LLT1 complex in the primary binding mode, and KACL residues contacting NKp65 in the NKp65:KACL complex, are highlighted in green. Purple triangles indicate LLT1 residues that engage NKR-P1 in the NKR-P1:LLT1 complex in the secondary binding mode. The alignment was performed in Clustal Omega, and the graphics was prepared in ESPript 3.0.

Fig. 2. The structure of human NKR-P1 shows a unique dimerization interface.

a Ribbon diagram of the NKR-P1 CTLD. Secondary structure elements are labeled in different colors: helix α1 is red, helix α2 is yellow, and β-strands and loops are cyan. b Comparison between NKR-P1 dimers formed by the glycosylated (cyan), deglycosylated free (green), and LLT1-bound (blue) forms of NKR-P1. c Comparison between helices α1- and α2-centered dimerization of murine dectin-1 (10.2210/pdb2CL8/pdb, magenta) and human LLT1 (10.2210/pdb4QKI/pdb, green), respectively; helices α1 and α2 are shown in red and yellow. Structural alignments of dectin-1 and NKR-P1 homodimers and LLT1 and NKR-P1 homodimers, prepared by aligning only one monomer from each dimer, are shown on the right-hand side. Although the CTLD fold is conserved in each pair of the aligned monomers, the helix α1- and helix α2-centered dimers show inverse arrangement.

Fig. 3. Glycosylation affects the dimerization interface of human NKR-P1.

a Dimerization interface of human NKR-P1. Subunits of human NKR-P1 are shown as Cα-trace (blue and cyan), and the dimer contact residues are shown as sticks with carbon atoms colored in light blue (blue subunit) and orange (cyan subunit); for clarity, only the residues of the blue subunit are labeled. The first GlcNAc unit N-linked to Asn116 and the carbohydrate chain N-linked to Asn169, observable in the NKR-P1_glyco structure, are shown with carbon atoms colored yellow and green, respectively. b Top view of the dimerization interface. The NKR-P1 subunits surfaces are colored blue and cyan. The GlcNAc units bound to Asn116 are shown as sticks with carbon atoms in yellow. Contact residues between the GlcNAc bound to chain A, and the chain B, are shown in yellow, whereas contact residues between the GlcNAc bound to chain B, and the chain A, are shown in purple. Hydrogen bonds are shown as green-dashed lines with a detailed view on the right-hand side. c Mixed glycosylation states at the dimer interface in the NKR-P1_deglyco structure. The GlcNAc unit N-linked to Asn157 of chain A is modeled with an occupancy of 0.5, while the second GlcNAc unit present at Asn116 of chain B is not modeled. Contours of 2mF_o-DF_c (2.8σ, cyan) and mF_o-DF_c (1σ, green) electron density maps are shown. d Small hydrophobic core in the central part of the NKR-P1 dimerization interface (subunits colored as in (a)). The central residues are shown as spheres with carbon atoms in yellow. The carbon atoms of Ile168 residues (whose mutation decreases the ability of NKR-P1 to bind LLT1) are shown in orange.

Fig. 4. The structure of the NKR-P1:LLT1 complex shows two distinct binding modes.

a The overall organization of the complex crystal structure. The LLT1 dimer (green/lemon) contacts the NKR-P1 dimer, formed by the blue and cyan monomers. The second blue-cyan NKR-P1 dimer is related to the first by crystal symmetry. The cyan NKR-P1 monomer interacts with LLT1 in the primary interaction mode, whereas the blue NKR-P1 monomer engages LLT1 using the secondary interaction interface. A black asterisk marks the mutual accessory contact of NKR-P1 bound in primary and secondary mode. Additionally, the asymmetric unit of the crystal contains another NKR-P1 dimer (pink/magenta) lacking contact with LLT1. b Overall comparison of the structure of dimeric KACL (purple) in complex with two NKp65 monomers (red; 10.2210/pdb4IOP/pdb) and the structure of the LLT1 dimer (green/lemon) with the two interacting NKR-P1 molecules in the primary (cyan, left side) and secondary (blue, right side) binding modes. Comparison with only the primary or secondary NKR-P1:LLT1 interaction modes is highlighted in the lower section (both in a side view, using 90° y-axis rotation). c, d NKR-P1:LLT1 primary and secondary interaction interfaces. Contact residues within 5 Å distance are colored in yellow. Amino acids forming the four strongest contacts are highlighted in red for the primary or magenta for the secondary mode.

Fig. 5. SEC-SAXS analysis of NKR-P1:LLT1 shows higher-order complex formation.

Overlay of the size-exclusion chromatography profile (black line) and SAXS scattering signal (red line) for the NKR-P1:LLT1 equimolar mixture at a 15 mg/ml loading concentration. Both signals show two distinct peaks. For each collected SAXS frame, the radius of gyration was calculated using AUTORG (blue circles). For further analysis, six intervals of the SAXS data were selected and separately merged (frames 355–367, 368–378, 379–388, 389–399, 431–440, 481–491; denoted as columns with diagonal hatching). SAXS scattering curve (a), Kratky plot (b), and pair distance distribution function (c) for the intervals 389–399 and 481–491 are shown in the inset for data quality assessment of the merged data. All six merged data intervals were then individually analyzed by OLIGOMER (Supplementary Data 1 and Supplementary Fig. 5). Diffraction data from the SEC-SAXS experiment have been deposited (10.17632/268ww2m4j3.1).

Fig. 6. Soluble LLT1 affects NKR-P1 distribution on the cell surface.

NKR-P1 stable transfectants were incubated in the presence or absence of soluble LLT1 or LLT1^SIM, and the cell surface distribution of NKR-P1 was monitored by super-resolution microscopy. a Representative brightfield (BF) and dSTORM images of full-length NKR-P1 HEK293 stable transfectants on PLL-coated slides incubated without (black) or with LLT1 (red) or LLT1^SIM (green), fixed and stained with AlexaFluor® 647-labeled anti-NKR-P1 mAb; scale bars represent 5 µm. The 10 µm² regions (red boxes in dSTORM images) are magnified and shown with corresponding clusters of fluorescent events (CoE) maps and binary maps; scale bars represent 1 µm. b–g Analysis of full-length NKR-P1 distribution relative changes induced by the presence of its LLT1 or LLT1^SIM soluble ligands: average cluster of events area (b), average cluster of events diameters (c), size distributions of cluster of events diameters overlaid with Poisson distribution functions (d), average events per cluster of events (e), density of events detected per cluster of events (f) and total density of the detected events (g). In b, c and e, f, each plotted point represents the mean value obtained from the analysis of the total inner surface of a single cell. The box plot center represents the overall mean value, the bounds of the box represent the interquartile range, and the whiskers represent ±SD. Data are from n = 45 NKR-P1⁺ control cells and n = 41 or n = 47 NKR-P1⁺ cells incubated with LLT1 or LLT1^SIM in seven or four independent experiments, respectively. One-way ANOVA with Bonferroni correction, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns. not significant. Source data are provided as a Source Data file.

Fig. 7. Inhibitory effect of soluble LLT1 on the NK cell-mediated cytotoxicity.

NK cells from three different donors (blue, red, and green) and K562 target cells were incubated with PBS buffer only (negative control) or with the soluble LLT1 or LLT1^SIM proteins, both in 50 and 250 µM concentrations corresponding to the 1× and 5× K_D values, respectively, as analyzed by the AUC for the primary interaction mode (cf. Supplementary Fig. 4). Bar charts represent the means of live K562 cells in each condition with results of individual experiments plotted as empty circles, and the whiskers represent ±SD. When applicable, data were statistically evaluated by one-way ANOVA. Considering p < 0.05 as statistically significant, the inhibitory effect of soluble LLT1 significantly differs from LLT1^SIM, which does not differ from the control condition lacking any NKR-P1 ligand. Source data are provided as a Source Data file.

Fig. 8. Organization of the NKR-P1:LLT1 complexes on the cell surface.

a Representation of four adjacent asymmetric units within the NKR-P1:LLT1 complex crystal, excluding the additional unrelated NKR-P1 dimer. The NKR-P1 (blue and cyan) and LLT1 (green and lemon) dimers alternate in primary (cyan and green) and secondary (blue and lemon) interactions, forming a chain-like structure. The schematic depiction of this arrangement is shown in the inset with the same color code. The black and white triangles represent N-termini positions, pointing behind and in front of the display plane, respectively. b Depiction of the hypothetical arrangement of the chain-like structure upon contact of an NK cell (bottom, blue) with a target cell (top, green) showing the crystal structure of two NKR-P1 dimers (cyan and blue) interacting with two LLT1 dimers (green and lemon) in the primary (cyan and green) and secondary (blue and lemon) modes. The first three N-terminal residues in the structures are highlighted in red. The flexible stalk regions connecting the N-termini and cell membranes are represented as speckled lines of the corresponding color-coding. The view on the right-hand side is clipped for clarity at the plane indicated on the left-hand side view. c Schematic depiction of NKR-P1 extracellular domain dynamics and possible ligand binding arrangements. NKR-P1 is expressed as a disulfide-linked homodimer; however, its CTLDs may undergo conformation change similar to monomer-dimer equilibrium. Such putative equilibrium would be shifted towards monomeric species for the wild-type protein and its I168T allelic variant. At the same time, a dimeric arrangement corresponding to the non-covalent dimer observed in the herein described crystal structures would be promoted for the S159A variant (left-hand side). Such NKR-P1 dimer could then interact with the cognate LLT1 ligand (itself being expressed as a disulfide-linked homodimer as well and forming stable non-covalent dimers with its CTLDs) in the previously suggested standard model of NK cell receptor – CTL ligand interaction (middle) or alternate with the dimeric ligand in the proposed chain-like arrangement based on NKR-P1:LLT1 complex crystal structure (right-hand side).