Ethical statement
Erythrocytes and human serum were obtained from the Japanese Red Cross (research ID 25J0143) with written informed consent. Parasites were collected from Thai patients according to the ethical approval of Chiang Mai University (permission number: 187/2554) and Mie University (permission number: 1312), and their use was approved by the University of Osaka (permission number: 149-003).
Cells and antibodies
The NKL cells were generously provided by L. L. Lanier at the University of California, San Francisco. The human erythroleukemia cell line (K562) was sourced from the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University. Anti-KIR2DL1/DS1 (Miltenyi Biotec; 130-118-973) and anti-KIR2DL1/DS5 (R&D Systems; MAB1844-SP) antibodies were purchased and used to detect KIR2DL1+ NK and KIR2DS1+ NK cells in PBMCs. The PBMCs were obtained from healthy donors. Anti-FLAG antibody (Sigma-Aldrich; F1804) and APC-conjugated anti-human IgG Fc antibody (Jackson ImmunoResearch; 109-136-098) were used to detect transgenic K562 and for binding assay involving Fc fusion proteins, respectively. Pacific Blue-conjugated anti-human CD107a (BioLegend; 328623), fluorescein isothiocyanate (FITC)-conjugated CD56 (BioLegend; 318303), PerCP/Cy5.5-conjugated anti-human IFNγ (BioLegend; 506527), APC/Cy7-conjugated anti-human TNF (BioLegend; 502943) and APC-conjugated anti-mouse-CD45 (BioLegend; 103111) antibodies were purchased and used for functional assay of RBK21.
Parasite culture
P. falciparum was cultured with human RBCs (type O blood; haematocrit 2%) in complete medium. This medium consisted of RPMI 1640 medium containing 10% human serum and 10% AlbuMAX I (Invitrogen), along with 25 mM HEPES, 0.225% sodium bicarbonate and 0.38 mM hypoxanthine. The medium was supplemented with 10 mg ml−1 of gentamicin. The cultures were maintained in an atmosphere containing 90% N2, 5% CO2 and 5% O2. Parasites were routinely synchronized every 4 days using a 5% sorbitol solution and subsequently used for analyses.
Cloning of RBK21 cDNA
The iRBCs that bound to KIR2DL1-Fc were enriched using the SH800 cell sorter (Sony). Enriched iRBCs were cultured, and total RNA was purified using TRIzol (Thermo Fisher Scientific). Subsequently, cDNA was synthesized using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific) and random hexamers according to the manufacturer’s instructions. The variable regions of RIFIN expressed in enriched iRBCs were amplified using the primers listed in Supplementary Table 3. The obtained fragment, which encoded a part of the RBK21 gene, was cloned into the centromere plasmid35 (pFCEN-rif), resulting in a chimaeric RIFIN. Expression was controlled by the promoter of elongation factor α of Plasmodium berghei. Before cloning the full-length RBK21 cDNA, the 5′-end and 3′-end of RBK21 cDNA were analysed using 5′-Full and 3′-Full RACE Core Sets (Takara) with specific primers (Supplementary Table 3). Following sequencing of the amplified product, a DNA fragment encompassing the entire coding region of RBK21 was obtained by means of polymerase chain reaction (PCR). The amplified fragment was cloned in the pFCEN-rif and used for further analysis.
Transfection of P.
falciparum
Transfection of P. falciparum was performed, as described previously36. In brief, schizonts of the P. falciparum 3D7 strain were purified using a Percoll/sorbitol gradient solution and cultured with fresh RBCs for 4 h. Subsequently, the parasites were synchronized within a 4-h window by treatment with 5% sorbitol. Full mature schizonts were purified from highly synchronized parasites and then transfected with pFCEN-rif plasmids using an Amaxa Nucleofector 2 with Parasite Nucleofector Kit 2 (Lonza). Because pFCEN-rif has human dihydrofolate reductase as a drug-selectable marker, transgenic parasites could be selected by drug treatment with pyrimethamine. Transfection experiments for each plasmid were carried out in duplicate throughout this study, resulting in biologically independent transgenic parasites.
Production of KIR-Fc proteins for library screening
The coding regions of KIR receptors KIR2DL1*00302 (Gene ID 3802; amino acid residues 22–245), KIR2DS1*001 (Gene ID 3806; amino acid residues 22–245), KIR2DL2*001 (Gene ID 3803; amino acid residues 22–245), KIR2DL3*001 (Gene ID 3804; amino acid residues 22–245), KIR2DL5*001 (Gene ID Q14953; amino acid residues 22–239), KIR3DL1*001 (Gene ID 3811; amino acid residues 20–339), KIR3DL2*001 (Gene ID 3812; amino acid residues 22–341) and KIR3DL3*001 (UniProt ID Q8N743; amino acid residues 25–320) were cloned into the vector pCAGSS. The resultant plasmids were introduced into HEK293T cell using TransIT-293 (Mirus Bio), and these KIRs were expressed as secreted human-Fc fusion proteins. These KIRs-Fc fusion proteins were obtained from supernatants of transfected HEK293 cells. To validate the correct folding of KIR-Fc fusion proteins, we used an established flow cytometry-based bead assay37. In brief, anti-KIR3DL2 antibody (BioLegend; 389602) or KIR-Fc fusion proteins were conjugated to Dynabeads M-450 Tosylactivated (Invitrogen; 14013) according to the manufacturer’s instructions. KIR-Fc beads were then blocked with PBS + 1% w/v bovine serum albumin (BSA) for 10 min under agitation before staining with the following antibodies: FITC anti-KIR2DL1/DL5 (R&D Systems; FAB1844F), PE anti-KIR2DL2/DL3/DS2 (BioLegend; 312605), PE anti-KIR2DL5 (Miltenyi Biotec; 130-096-199), FITC anti-KIR3DL1 (BioLegend; 312705), anti-KIR3DL2 (BioLegend; 389602) and anti-KIR3DL3 (R&D Systems; FAB8919P). Anti-KIR3DL2 beads were blocked with PBS + 1% w/v BSA for 10 min before staining with KIR3DL2-hFc for 20 min under agitation. If secondary detection was required, the beads were washed twice with cold PBS before staining with either 2 µg ml−1 of AF647-labelled goat anti-human F(ab′)2 (Jackson ImmunoResearch; 709-605-098) or AF647-labelled donkey anti-mouse F(ab′)2 (Jackson ImmunoResearch; 715-606-150) for 20 min under agitation. The beads were then washed twice in cold PBS before analysis on BD LSR II using BD FACSDiva software.
Flow cytometry
The binding of KIRs-Fc fusion proteins to iRBCs was assessed by flow cytometric analysis using an Attune NxT (Thermo Fisher Scientific). Before the assay, the KIRs-Fc fusion proteins were mixed with an APC-conjugated anti-human IgG Fc antibody. The iRBCs were mixed with the APC-labelled KIRs-Fc binding proteins, and parasite nuclei were stained with Hoechst 33342. The iRBC populations were gated on the basis of fluorescence from Hoechst 33342, and KIR-Fc-bound iRBCs were selected using APC fluorescence (Supplementary Fig. 2). All assays were performed in triplicate, and all data were analysed using FlowJo software (Becton Dickinson).
The iRBCs binding to KIR2DL1 were selectively sorted using the SH800 cell sorter (Sony). Before sorting, iRBCs containing late trophozoites and schizonts were separated from uninfected RBCs using the Percoll/sorbitol gradient solution, followed by mixing with APC-labelled KIR2DL1-Fc fusion protein. After binding KIR2DL1-Fc to iRBC, they were suspended in complete medium and subjected to SH800. The sorted iRBCs were immediately recovered in complete medium and cultured with fresh RBCs.
Preparation of RIFIN expression library
DNA fragments that encoded all RIFIN variable regions were amplified from the genomic DNA of the 3D7 strain using degenerated primers designed at the internal sites of the N-terminal and C-terminal conserved regions of RIFINs (Supplementary Table 3). The amplified fragments representing the variable regions of all RIFINs of the 3D7 strain were cloned into the BsmBI sites of the pFCEN-rif plasmid (Supplementary Fig. 1). The cloned fragments were flanked with the N-terminal and C-terminal conserved regions of pFCEN-rif, resulting in chimaeric RIFIN genes. The pFCEN-rif plasmids containing chimaeric RIFIN genes were introduced into the P. falciparum 3D7 strain through single electroporation. Transgenic parasites were subsequently obtained by means of treatment with pyrimethamine, resulting in the creation of a RIFIN expression library. Each parasite within the RIFIN expression library expressed distinct chimaeric RIFINs. Library construction was performed twice, resulting in the establishment of two biologically independent RIFIN expression libraries, designated rif-lib1 and rif-lib2.
The RIFIN expression libraries from Lek174 and Lek79 were generated in a manner similar to that of the 3D7 strain. Briefly, DNA fragments encoding the RIFIN variable region were amplified using the degenerated primers (Supplementary Table 3), cloned into the pFCEN-rif plasmid and introduced into the 3D7 strain. Two biologically independent libraries were generated for each of the two field-isolated parasite strains and were used for screening using KIR2DL1-Fc.
Screening for KIR-binding RIFINs from expression libraries
The iRBCs of rif-lib1 and rif-lib2 were incubated with the KIR2DL1-Fc fusion protein and screened using the cell sorter. The cells were cultured in a complete medium immediately after sorting. To identify the selected RIFINs, the pFCEN-rif plasmids containing chimaeric RIFIN genes were recovered from the sorted iRBCs. The variable regions integrated into the plasmids were subsequently amplified by PCR using primers designed on the basis of the plasmid backbone sequence and recovered plasmid as a template (Supplementary Table 3). Following this, sequence tags were introduced at the ends of the amplified DNA fragments by means of PCR, and the resultant tagged DNA fragments were sequenced using MiSeq (Illumina). The reads containing the variable regions were isolated from the obtained FASTQ files using the SeqKit program38 with the ‘grep’ option. Subsequently, the plasmid backbone sequence was trimmed using the Trimmomatic 0.39-2 program39. After confirming the sequence quality of the trimmed reads, they were mapped onto the reference genome sequence of the P. falciparum 3D7 strain, which was downloaded from PlasmoDB 62 using the bowtie2 tool40. Reads aligned to more than two different genomic loci were eliminated from the mapping data. The number of reads for each RIFIN gene was tallied using featureCounts41 and normalized by dividing it by the total number of mapped reads. The normalized read count for each RIFIN gene was further divided by the normalized read count obtained before screening. Candidates for KIR2DL1-binding RIFINs were identified on the basis of values exceeding 2. Each variable region of a candidate RIFIN was amplified from the genomic DNA of the P. falciparum 3D7 strain and individually cloned into pFCEN-rif. Transgenic parasites were generated by transfecting the resultant plasmids into the 3D7 strain, and binding to KIR2DL1 was assessed using Attune NxT and Fc fusion protein.
The expression libraries for RIFINs from Lek174 and Lek79 were selected using a method similar to that used for 3D7. The variable regions of the candidate RIFINs were amplified from the recovered pFCEN-rif and tagged, followed by sequencing using MiSeq. The obtained sequence reads were processed using the SeqKit and Trimmomatic 0.39-2 program in a similar manner to those from rif-lib1 and rif-lib2. The genomes of Lek174 and Lek79 were sequenced using MinION (Oxford Nanopore Technologies) and MiSeq, and their genomic contigs were generated from the obtained long reads and short reads using Flye42, BWA43,44 and Pilon45 programs. The processed sequenced reads of variable regions were then mapped onto the generated contig sequences, and the sequence information for the region of the contig to which the reads were mapped was manually obtained, together with 1,000 bp upstream and downstream. The general feature format files were generated using the sequence information of the region to which reads were mapped and contig ID, and the number of mapped reads was counted using featureCounts 2.0.1 program, followed by ranking the regions containing mapped reads on the basis of the number of reads. The top 20 regions were selected, and ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder/) was used to analyse these regions and identify the variable regions of the KIR2DL1-binding RIFN candidates. All the experiments were conducted using two biologically independent libraries for each of two strains.
Phylogenetic analysis of RIFINs
The amino acid sequences of all RIFINs of the P. falciparum 3D7 strain were obtained from PlasmoDB (https://plasmodb.org/plasmo/app/) and aligned using the Clustal Omega program (https://www.ebi.ac.uk/jdispatcher/msa/clustalo). Regions equivalent to the variable region of the LILRB1-binding RIFN were selected (PF3D7_1254800)3. These were aligned using Clustal Omega, followed by the generation of a phylogenetic tree on the basis of the newly aligned sequences. All analyses were performed with default setting. The resultant data from the phylogenetic analysis were visualized using the TreeView program46. The amino acid sequences of the variable regions of KIR2DL1-binding RIFIN candidates from Lek174 and Lek79 were obtained as described above and analysed together with the variable regions of all RIFINs from the 3D7 strain. For prediction of KIR2DL1-binding RIFIN clades in field isolate genomes, 3D7 KIR2DL1-binding RIFIN sequences were combined with the RIFIN repertoire of either PfKE01 or PfSN01, as obtained using PlasmoDB, and the same multiple sequence alignment approach was used, as described above.
Transfection of mammalian cells
Stable transfectants of K562–RBK21, K562-negative control rifin (PF3D7_1254200), NKL–KIR2DL1 and NKL–KIR2DL3 were generated using retrovirus-mediated transduction with the pMXs retroviral expression vector and PLAT-E retroviral packaging cells transfected with the amphotropic envelope, as described previously2,47 (Cell Biolabs). Briefly, the variable regions of these RIFINs were fused with the transmembrane and cytoplasmic regions of PILRα (amino acid residues 196–256) and cloned into the pMXs plasmid. The resultant plasmid containing the fusion gene of RBK21 and PILRα was introduced into PLAT-E cells using TransIT-293 (Mirus Bio), and the recombinant retrovirus was collected from the supernatant 3 days after transfection. The full-length cDNA of KIR2DL1 and KIR2DL3 was chemically synthesized (GenScript) and subsequently cloned into the pMXs plasmid. Recombinant retrovirus was then produced using a method similar to that described above. The target cells (K562 and NKL) were seeded in a 24-well culture plate and infected with the produced recombinant virus, resulting in K562–RBK21, K562-negative control rifin, NKL–KIR2DL1 and NKL–KIR2DL3.
Reporter assay
The extracellular domains of KIR2DL1 and KIR2DS1 were fused with PILRβ and expressed on the surface of mouse T cell hybridomas that were stably transfected with NFAT–GFP and FLAG-tagged DAP12, as described previously48. The transmembrane domain of PILRβ can transduce the signal through the DAP12 adaptor molecule in the reporter cell, resulting in the expression of GFP. The reporter cell lines, which expressed fusion proteins of KIR2DL1 and KIR2DS1 with PILRβ, were stimulated by iRBCs, which were infected with parasites expressing RBK21 for 16 h, and their GFP expression, which was selected using anti-CD45 antibody, was monitored using Attune NxT. The P. falciparum 3D7 strain was used as a negative control.
Measurement of CD107a, IFNγ and TNF in NK cells
NK cells were purified using an NK cell isolation kit (Miltenyi Biotec), according to the manufacturer’s instructions, from PBMCs obtained from two donors positive for KIR2DL1 and two donors positive for KIR2DS1. Following purification, the NK cells were cultured in the NK cell growth medium, which consisted of RPMI 1640 supplemented with 15% FCS (Gibco), 5% human serum, 1× minimum essential medium non-essential amino acids, 1 mM sodium pyruvate, 100 U ml−1 of penicillin, 100 μg ml−1 of streptomycin, 500 U ml−1 of human interleukin (hIL)-2, 5 ng ml−1 of hIL-15, 10 ng ml−1 of hIL-12, 40 ng ml−1 of hIL-18 and 20 ng ml−1 of hIL-21. To evaluate the inhibitory effect of RBK21 on NK cell function, purified NK cells (1.0 × 105 cells) containing KIR2DL1-positive NK cells were co-cultured with K562–RBK21 cells (1.0 × 105 cells) in a 96-well plate at 37 °C for 4 h. Pacific Blue-conjugated anti-human CD107a antibody was added to the medium at the beginning of co-culturing for the measurement of CD107a expression. After 1 h of co-culture, the NK cells were treated with BD GolgiStop reagent, which is included in the BD Cytofix/Cytoperm Fixation/Permeabilization Kit containing monensin (BD Biosciences). The NK cells were collected by centrifugation, and dead cells were stained at room temperature for 30 min using a LIVE/DEAD Fixable Yellow Dead Cell Stain Kit (Invitrogen). The NK cells were further stained on ice for 30 min with FITC-conjugated anti-CD56 antibody, PE-labelled anti-KIR2DL1/DS1 antibody and APC-labelled anti-KIR2DL1/DS5 antibody. The cells were fixed and permeabilized at 4 °C for 20 min with a fixation/permeabilization solution, which was also included in the kit. The permeabilized NK cells were then stained at 4 °C for 30 min in the dark with PerCP/Cy5.5-conjugated anti-human IFNγ antibody and APC/Cy7-conjugated anti-human TNF antibody using the BD Perm/Wash buffer. CD107a expression and production of IFNγ and TNF were assessed using Attune NxT. The K562 and K562-negative control RIFIN (PF3D7_1254200) were used as negative controls. To assess the activation of NK cells through KIR2DS1 by RBK21, NK cells containing KIR2DS1-positive NK cells were co-cultured with RBCs infected with RBK21-expressing transgenic parasites. The 3D7 strain and transgenic parasites expressing a negative control RIFIN were used as negative controls. CD107a expression and IFNγ and TNF production were assessed, as described above. All assays were performed in six or nine replicates, and representative raw flow cytometry data are shown in Extended Data Figs. 5 and 8.
Cytotoxicity assay
Suppression of the cytotoxic activity of NKL–KIR2DL1 cells by RBK21 was examined, as described previously2. The viabilities of K562–RBK21 and NKL–KIR2DL1 were assessed before the assay. If the viability was lower than 85%, the cells were purified using Ficoll and cultured for 3 days. The highly viable K562–RBK21 was labelled with 15 μM calcein acetoxymethyl ester (calcein-AM) in assay medium (RPMI 1640 without phenol red, supplemented with 1% FCS) for 30 min at 37 °C, followed by washing twice with the assay medium. The NKL–KIR2DL1 cells were washed twice with the assay medium and mixed with K562–RBK21 in a 96-well plate with effector (NKL–KIR2DL1) to target (K562–RBK21) ratios ranging from 1:1 to 1:8 in triplicate. These cells were centrifuged at 250g for 5 min and then co-cultured for 4 h at 37 °C. To detect the maximal release of calcein-AM from K562–RBK21, cells were dispensed without NKL–KIR2DL1 in the plate, followed by lysis with 1% Triton X-100 for 30 min before the end of co-incubation. In addition, the cells were cultured without NKL_KIR2DL1, and the spontaneous release of calcein-AM in the supernatant was measured. After co-incubation, the cells were centrifuged at 1,500 rpm for 2 min, and the fluorescence of the released calcein-AM was measured. Specific cytotoxicity C (%) was calculated using the following formula: C = 100 × (mean fluorescence in co-culture − mean fluorescence in spontaneous lysis)/(mean fluorescence in maximal lysis − mean fluorescence in spontaneous lysis). K562 and K562-negative control RIFIN (PF3D7_1254200) were used as positive controls. In addition, NKL and NKL–KIR2DL3 were tested. All assays were performed in quadruplicate.
Nanobody identification and screening
To identify KIR2DL1-binding nanobodies, a complex of KIR2DL1 (residues 27–221) and RBK21 variable domain (residues 148–299) was produced and purified by size exclusion chromatography using a Superdex 75 10/300 column (Cytiva). After three rounds of llama immunization, a VHH cDNA library was generated, and nanobodies were screened in a phage–enzyme-linked immunosorbent assay approach using a biotinylated complex, adapted from a previous study49. Hits from the enzyme-linked immunosorbent assay screen were sequenced, and non-redundant nanobody sequences were expressed using periplasmic expression in a WK6 Escherichia coli strain. The VHH domain was flanked by an N-terminal PelB leader sequence and a C-terminal His6-tag in a pET15b vector, and the nanobodies were expressed and purified, as detailed in a previous study49. The nanobodies were then screened by means of SPR, and the nanobody with the most favourable kinetic profile was selected for grafting onto a NabFab nanobody scaffold50. The resultant nanobody (Nb1) was expressed as previously but with the addition of a final polishing step using a Superdex 75 10/300 column (Cytiva).
Protein expression and purification
To produce proteins for structural biology, the coding sequence for KIR2DL1 (residues 27–221) was cloned into the vector pHL-sec, including a C-terminal His6-tag, and transfected into Mycoplasma-screened Expi293F GnTI- cells using ExpiFectamine 293 (Thermo Fisher Scientific). Mutagenesis of RBK21 constructs was carried out using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs), following the manufacturer’s instructions. Five days after transfection, the supernatant was collected by centrifugation at 5,000g to pellet the cells before filtering with a bottle-top 0.45 µM filter. The pH was adjusted to 7.5 through the addition of Tris to a final concentration of 50 mM before passing over the Ni Sepharose excel resin (Cytiva). The captured KIR2DL1 protein was washed with 50 mM Tris (pH 7.5), 150 mM NaCl, 20 mM imidazole before elution with 50 mM Tris (pH 7.5), 150 mM NaCl and 500 mM imidazole. The protein was incubated with Endo Hf overnight at 37 °C to cleave glycans.
The RIFIN variable domain constructs (RBK21 residues 148–299 and KEN-01 residues 159–288 with A165C and A282C mutations) were cloned into a modified pENTR4LP vector, including a cleavable N-terminal monomeric Fc domain tag and a C-terminal C-tag before transfection into Expi293F GnTI- cells, as described above. The supernatant was processed, as described above, and flowed over the CaptureSelect C-tagXL Affinity Matrix (Thermo Fisher Scientific) to isolate the C-tagged protein. The captured RBK21-mFc was washed with 50 mM Tris (pH 7.5) and 150 mM NaCl and eluted with 2 M MgCl2 and 20 mM Tris (pH 7.5) before buffer exchange with 50 mM Tris (pH 7.5) and 150 mM NaCl. The protein was incubated with Endo Hf and tobacco etch virus protease overnight at 37 °C to cleave glycans while simultaneously releasing the variable domain from the mFc tag. The RIFIN was flowed over Pierce Protein G Agarose to remove the mFc fusion tag, and the flow-through was collected before further purification by size exclusion chromatography using a Superdex 75 10/300 column (Cytiva).
The HLA-Cw4–β2m complex used in the SPR experiments was produced using a refolding protocol, as described previously51. In brief, HLA (accession no. MH254935) and β-2-microglobulin (β2M; accession no. NM_004048) were expressed separately in E. coli BL21(DE3) pLysS and purified as inclusion bodies. These were solubilized in a buffer containing 6 M guanidine HCl, 100 mM Tris-HCl (pH 8.0), 2 mM EDTA and 0.1 mM dithiothreitol. Solubilized HLA and β2m were diluted to 10 mg ml−1 using solubilization buffer, with the addition of 1 M dithiothreitol to a final concentration of 10 mM, and the solutions were incubated at room temperature for 1 h. Refolding buffer (100 mM Tris (pH 8.0), 400 mM l-Arg, 2 mM EDTA, 3.73 mM cystamine and 6.73 mM cysteamine) was chilled to 4 °C before β2m was refolded by rapid dilution, to a final concentration of 2 µM. Refolding was allowed to proceed for 2 h before the addition of 10 mg l−1 of peptide (QYDDAVYKL), and then HLA C was added dropwise to a final concentration of 2 µM. The heterotrimer was allowed to refold for 72 h at 4° before size exclusion chromatography (HiLoad 26/600 Superdex 75 pg) in 20 mM Tris-HCl (pH 8.0) and 100 mM NaCl.
Crystallization, data collection and structure determination
KIR2DL1 and RBK21 were combined to a 1:1 molar ratio, and the resulting complex was purified using a Superdex 75 10/300 column (Cytiva) in 20 mM HEPES and 150 mM NaCl. Initial crystallization trials were carried out using vapour diffusion in sitting drops by mixing 100 nl of protein solution (10 mg ml−1) with 100 nl of well solution using commercial screens. For RBK21–KIR2DL1, initial hits were obtained after 10 days at 18 °C in condition H12 (0.1 M Tris-bicine (pH 8.5), 0.1 M amino acid mix and 37.5 % v/v precipitant mix 4) of the Morpheus HT-96 screen (Molecular Dimensions). Using these initial crystals as seeds, an optimization screen was set up using the Morpheus stock solutions, screening a pH range from 8.0 to 9.0 and a final precipitant concentration between 21% and 28% by mixing 100 nl of protein solution with 100 nl of well solution and 25 nl of seed stock. The best crystals were obtained after 12 days with a well solution of 0.1 M Tris-bicine, 0.1 M amino acid mix and 23% precipitant mix 4 and were collected then cryo-cooled for data collection in liquid nitrogen.
Data were collected on beamline ID30A-3 at ESRF at a wavelength of 0.97625 Å, indexed using DIALS and scaled using AIMLESS, resulting in a full dataset with a final resolution of 2.89 Å. The structure was solved by means of molecular replacement using KIR2DL1 (PDB 1IM9) as a search model using Phaser-MR (v.2.8.3)52. Building and refinement cycles were carried out using Coot (v.0.8.9.2)53 and BUSTER (v.2.10)54.
For KEN-01, a complex was formed using a 1:1.2:1.5 ratio of KEN-01:KIR2DL1:Nb1, and an equimolar complex was obtained through purification on a Superdex 75 10/300 column (Cytiva) in 10 mM HEPES and 75 mM NaCl. The same preliminary screening strategy was implemented as described above, and crystals were obtained after 12 days at 18 °C in condition E3 of the PEG/ION-HT screen (Hampton Research; 0.1 M sodium malonate (pH 5.0) and 12% PEG 3350). These crystals were collected in mother liquor + 25% glycerol before cryo-cooling in liquid nitrogen. Data were collected on beamline i03 using a Diamond Light Source at a wavelength of 0.976246 Å. Indexing and data reduction were performed using xia2 DIALS, resulting in a full dataset with a final resolution of 2.17 Å. The previously solved KIR2DL1 and RBK21 structures were used as molecular replacement search models to solve the new dataset (Phaser-MR v.2.8.3 (ref. 52)) and preceded cycles of building and refinement using Coot and BUSTER.
Surface plasmon resonance
All experiments were conducted on a Biacore T200 instrument (Cytiva) using a running buffer of 20 mM HEPES (pH 7.5), 300 mM NaCl and 0.005% v/v Tween-20. Proteins were desalted in this buffer using PD-10 columns (Cytiva). Sensorgrams were double referenced by subtraction of the response measured from a blank flow path with no protein immobilized, in addition to subtraction of the response attributable to buffer from the protein flow path. Kinetic values were obtained using the BIAevaluation software (Cytiva) by fitting data to a global 1:1 interaction model, allowing for the determination of the association rate constant (kon), dissociation rate constant (koff) and affinity (KD). Equilibrium fits of the multicycle experiments were obtained using a 1:1 interaction model in BIAevaluation (Cytiva). All kinetic and equilibrium fits are contained within the Source Data.
For experiments comparing its binding affinity for KIR2DL1 and KIR2DS1, RBK21 was coupled to a CM5 sensor (Cytiva) through amine chemistry at approximately 150 response units (RU). A twofold dilution series of KIR2DL1 and KIR2DS1 was injected over the chip for 240 s at a flow rate of 30 µl min−1, followed by a 300-s dissociation time and regeneration between cycles with 10 mM glycine (pH 2.5) for 10 s. Affinity values were derived from both equilibrium and kinetic fits to the data.
To measure HLA-Cw4 binding, approximately 300 RU of each Fc-tagged KIR2DL1 and KIR2DS1 was captured through Protein A/G (Thermo Fisher Scientific) pre-immobilized on a CM5 sensor (approximately 1,500 RU on each flow cell). The same association and dissociation times as described above were used. Regeneration of the surface was carried out with 10 mM glycine (pH 2.5) before more KIR2DL1 and DS1 Fc-tagged protein were re-immobilized on the chip, and the next cycle was started. Affinity values for these data could only be fit using equilibrium measurements because of fast on-rate and off-rate, which make kinetic fits unreliable.
Comparison of the binding of wild-type and S221R RBK21 to KIR2DL1 was achieved through immobilization of approximately 300 RU of KIR2DL1 onto the surface of a CM5 sensor followed by injection of a twofold dilution series of RBK21 variants starting from 20 µM over the sensor surface at a flow rate of 30 µl min−1. The association and dissociation times were 60 and 120 s, respectively, with regeneration with 10 mM glycine (pH 2.5) for 10 s between cycles.
RIFINs from field-isolated strains were screened for KIR2DL1/DS1 binding through immobilization of KIR2DL1 and KIR2DS1 and a negative control of LILRB1 to a CM5 chip sensor (Cytiva) using amine chemistry (500, 500 and 1,000 RU immobilized, respectively). The mFc-tagged RIFINs were exchanged into a running buffer of 20 mM HEPES (pH 7.5), 300 mM NaCl and 0.005% v/v Tween-20 and flowed over the sensor in multicycle experiments, as described above.
For competition experiments, a paired test was performed under saturating conditions. KEN-01 was flowed over the sensor either alone (5 µM) or mixed with another tested protein (both proteins at 5 µM). Response was measured 5 s before the dissociation phase, and all responses were expressed as a percentage increase relative to 5 µM of KEN-01 alone. A positive control of KEN-01 and Nb1 demonstrated how a known non-competitive binder resulted in a marked increase in response, whereas RBK21, a known competitive binder, showed no significant increase. Paired tests were performed with all field-isolated RIFINs in triplicate.
Circular dichroism
Proteins were desalted into 10 mM sodium phosphate (pH 7.5) and 100 mM NaF using PD-10 columns (Cytiva). Measurements were made on a JASCO J815 CD spectrophotometer. Experiments were recorded at 20 °C between 190 and 260 nm at 0.5-nm intervals with a protein concentration of 0.1 mg ml−1 using a 1-mm path-length cuvette (Hellma Macro Cell 110-QS). Three equivalent protein spectra were recorded and averaged after subtraction of a buffer-only blank measurement. Data were processed using the CAPITO online web server55.
Shannon entropy calculation
The variable region from each KIR2DL1-binding RIFIN identified in the clade was used to generate a multiple sequence alignment using MUSCLE56. The multiple sequence alignment was then provided to the Protein Variability Server57 to calculate entropies. Per-residue entropy values were then binned into three categories: absolute conservation, high conservation and medium conservation with entropy values of 0, 0–0.5 and 0.5–0.8, respectively. These three bins were mapped onto the structure in PyMOL.
Supported lipid bilayer assay
SLB experiments were performed, as described previously3. Briefly, 1,2-dioleoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids) micelles were supplemented with 12.5% 1,2-dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl) iminodiacetic acid) succinyl]-Ni (Avanti Polar Lipids) and infused into plasma-cleaned glass coverslips affixed within a six-lane adhesive chamber (ibidi) for 20 min. SLBs were washed three times with HEPES buffered saline + 0.1% BSA + 1 mM CaCl2 + 2 mM MgCl2 and blocked with 100 μM NiSO4 in 5% BSA/PBS. After another washing step, protein dilutions were added to achieve 600 molecules µm−2 ICAM-1 and 100 molecules µm−2 PfRH5, with or without 100 molecules µm−2 of the indicated RIFIN, as determined by flow cytometry with bead-supported bilayers. The protein mixtures were incubated for 20 min to allow for attachment and then washed. Monoclonal antibody R5.016 was added at 2 μg ml−1 for 20 min followed by washing. Then, 106 NK cells, isolated from fresh blood samples using a RosetteSep Human NK Cell Enrichment Cocktail (STEMCELL Technologies), were infused into each lane, followed by incubation for 30 min at 37 °C. The bilayers were then fixed for 5 min in 4% paraformaldehyde/Hank’s balanced salt solution, followed by washing. Perforin staining was performed using monoclonal anti-perforin Alexa Fluor 488 (clone dG9; BioLegend) at a concentration of 10 μg ml−1 for at least 1 h. The bilayers were washed three times before image acquisition.
Imaging was performed, as described previously3, using an Olympus cell TIRF-4Line system with a ×150 (numerical aperture: 1.45) oil objective at room temperature. We analysed images using ImageJ (v.1.54b; National Institutes of Health). Cell boundaries were defined on the basis of segmented (‘default’ algorithm in ImageJ) interference reflection images58.
Homology modelling of KIR2DS1
The SWISS-MODEL59 web interface was used to generate a template homology model of KIR2DS1 for structural comparisons. The protein sequence coding for domains D1 and D2 of KIR2DS1 was inputted into the tool with a PDB template supplied by KIR2DL1 from our crystallographic data.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
