Peptide-specific engagement of the activating NK cell receptor KIR2DS1

The activating NK cell receptor KIR2DS1 has been shown to be involved in many disorders including autoimmune diseases, malignancies and pregnancy outcomes. However, the precise ligands and functions of this receptor remain unclear. We aimed to gain a better understanding of the factors involved in the binding of KIR2DS1 and its inhibitory counterpart KIR2DL1 to HLA class I molecules, and the consequences for KIR2DS1+ NK-cell function. A systematic screen that assessed binding to 97 HLA-I proteins confirmed that KIR2DS1-binding was narrowly restricted to HLA-C group 2 complexes, while KIR2DL1 showed a broader binding specificity. Using KIR2DS1ζ+ Jurkat reporter-cells and peptide-pulsed 721.221.TAP1KO-HLA-C*06:02 cells, we identified the synthetic peptide SRGPVHHLL presented by HLA-C*06:02 that strongly engaged KIR2DS1- and KIR2DL1-binding. Functional analysis showed that this HLA-C*06:02-presented peptide can furthermore activate primary KIR2DS1(+) NK cell clones. Thus, we demonstrated peptide-dependent binding of the activating NK cell receptor KIR2DS1, providing new insights into the underlying mechanisms involved in KIR2DS1-related disorders.

KIR2DS1 and KIR2DL1 binding to HLA-C2 molecules is peptide-dependent. Several studies have revealed that peptides presented by HLA class I molecules can influence KIR binding [19][20][21]25 . We therefore investigated whether specific peptides presented by the HLA-C2 allele HLA-C*06:02 can impact KIR2DS1-as well as KIR2DL1-binding. To prevent 721.221-HLA-C*06:02 cells from presenting self-peptides onto HLA-C*06:02, 721.221-HLA-C*06:02 cells with a knock-out for TAP1 were produced. This allowed for the controlled identification of externally added peptides that bind to and stabilize HLA-C*06:02 expression on the cell surface. HLA stabilization was quantified by measuring HLA expression on 721.221.TAP1KO-HLA-C*06:02 cells using flow cytometry.
We first tested 19 synthetic peptides that had been previously described to bind to HLA-C*06:02 26 . Furthermore, 568 overlapping peptides spanning the entire HIV-1 clade B sequence peptides (346 18aa-long peptides covering the entire HIV-1 consensus sequence and 222 decametric peptides overlapping by 9 amino acid and covering p24 GAG) were assessed for their potential to stabilize HLA-C*06:02. Peptides inducing a robust increase of HLA-C*06:02-expression (MdFI higher than 2 S.D. above the mean of non-stabilizing control peptide (LLRHHNLIY)) were defined as HLA-C*06:02-binding peptides, resulting in the identification of 20 peptides presented by HLA-C*06:02 (Fig. 3a). The peptides included 14 of the 19 previously described synthetic HLA-C*06:02-binding peptides and 6 novel HIV-1-derived peptides. Most of the peptides stabilizing HLA-C*06:02 showed the same binding motif consisting of a phenylalanine in position 1 of the peptide sequence (15/20), an arginine in position 2 of the peptide sequence (15/20) and an aliphatic amino acid (valine, leucine or isoleucine) in position 9 of the peptide sequence (16/20), which is in agreement with the previously defined binding motif for HLA-C*06:02 [26][27][28] (see Supplementary Table 1).
The 20 selected peptides that stabilized HLA-C*06:02 expression were subsequently tested for their ability to engage KIR2DS1-and KIR2DL1 by co-incubating KIR2DS1ζ + and KIR2DL1ζ + Jurkat reporter cells with 721.221-TAP1KO-HLA-C*06:02 cells loaded with the respective peptides (Fig. 3b). Only one out of the 20 tested peptides, SRGPVHHLL (HLA-Cw6-SV9), showed a significant increase of reporter cell activity for both KIR2DS1ζ + and KIR2DL1ζ + Jurkat cell lines. None of the 6 identified HLA-C*06:02-binding HIV-1 peptides induced any activation of Jurkat reporter cells lines. The SRGPVHHLL peptide furthermore increased the reporter cell activity of both KIR2DS1ζ + and KIR2DL1ζ + Jurkat cells in response to TAP-competent 721.221. HLA-C*06:02 cells (Fig. 3c). Blocking experiments (Fig. 3d) were performed by pre-incubation of 721.221. TAP1KO-HLA-C*06:02 cells pulsed with the peptide SRGPVHHLL with an HLA-C blocking antibody (clone 6A4, IgM) before incubation with KIR2DS1ζ + and KIR2DL1ζ + Jurkat cells and resulted in the abrogation of reporter cell activity, confirming that KIR2DS1ζ + and KIR2DL1ζ + Jurkat cells activation was HLA/ peptide-dependent.
To further confirm our findings, HLA-C*06:02 tetramers folded with the peptide SRGPVHHLL and conjugated to the fluorescent molecule PE (referred as HLA-C*06:02-SRGPVHHLL-PE) were used to stain the KIR2DS1ζ + and KIR2DL1ζ + Jurkat cells (Fig. 3e). A strong increase of PE MdFI signal was observed when the KIR2DS1ζ + and KIR2DL1ζ + Jurkat cells were stained with the HLA-C*06:02-SRGPVHHLL-PE tetramer. Taken together, out of the 587 peptides tested, we identified one synthetic peptide, SRGPVHHLL, which stabilized HLA-C*06:02-expression and induced strong and functionally relevant binding of KIR2DS1 and KIR2DL1.
Amino acid variations within the HLA-C*06:02-restricted SRGPVHHLL peptide impact KIR2DS1 binding. To determine the peptide residue important for modulating KIR2DS1-and KIR2DL1-binding, various amino acid substitutions were introduced in the SRGPVHHLL peptide sequence. A panel of 8 peptides (see Supplementary Table 2) was synthetized with amino acid substitutions at P2 or P9, identified previously as anchor residues for HLA-C*06:02-binding [26][27][28] , or P7, a position described as important for KIR recognition [19][20][21] . The amino acid substitutions were selected to cover the principal categories of amino acids according to their side chain (polar, apolar, neutral, acidic, basic). Peptides were first tested for HLA-C*06:02 stabilization and 7 out of the 8 peptides were identified to stabilize HLA-C*06:02-expression, demonstrating that most single amino acid substitution did not affect HLA-C*06:02-binding (Fig. 4a). The peptide R 2 A/L 9 A did not show any stabilization of HLA-C*06:02, suggesting that amino acid substitutions in both anchor position (P2 and P9) abrogated HLA-C*06:02-binding. Subsequently, the peptide variants were tested for their ability to modify KIR2DS1ζ +  and KIR2DL1ζ + binding when presented by HLA-C*06:02 ( Fig. 4b and c). The peptide R 2 A/L 9 A, which did not stabilize HLA-C*06:02, was used as an additional negative control. The peptides R 2 A and L 9 A, which stabilized HLA-C*06:02-expression, decreased both KIR2DS1ζ + and KIR2DL1ζ + Jurkat cell activity to the same level as the negative control peptide (LLRHHNLIY), showing that single substitution in P2 or P9 abrogated KIR2DS1 as well as KIR2DL1 binding. The peptides with the sequence changes H 7 A, H 7 G and H 7 S did not affect KIR2DS1ζ + and KIR2DL1ζ + Jurkat cell reporter activity compared to wild type peptide. Finally, the peptide with the H 7 R substitution slightly decreased the reporter cell activity of KIR2DS1ζ + Jurkat cells, but not the activity of KIR2DL1ζ + Jurkat cells, while the peptides with the H 7 D and H 7 S substitutions slightly decreased the reporter cell activity of KIR2DL1ζ + Jurkat cells, but not of KIR2DS1ζ + Jurkat cells. In summary, we demonstrated that binding of KIR2DS1 and KIR2DL1 to the SRGPVHHLL peptide presented by HLA-C*06:02 was modulated by the peptide sequence.
SRGPVHHLL presented by HLA-C*06:02 triggers degranulation of primary KIR2DS1+NK cell clones. The functional consequences of peptide-dependent engagement of activating KIRs remains insufficiently understood. The effects of the SRGPVHHLL peptide presented by HLA-C*06:02 on KIR2DS1 binding was therefore investigated by performing NK cell degranulation assays using primary KIR2DS1(+) KIR2DL1(−) and KIR2DS1(−) KIR2DL1(−) NK cell clones derived from KIR2DS1+ HLA-C1/C1 individuals (see Supplementary Fig. 1 for gating strategy and characterization of NK cells clones). As displayed in Fig. 5a and b, 721.221-TAP1KO-HLA-C*06:02 cells pulsed with the SRGPVHHLL peptide induced strong degranulation of KIR2DS1(+) KIR2DL1(−) NK cell clones, as measured by the percentage of CD107a(+) NK cells, compared to 721.221.TAP1KO-HLA-C*06:02 cell alone or pulsed with the control peptide LLRHHNLIY. In contrast, KIR2DS1/KIR2DL1 double-negative NK cell clones were not activated by the peptide SRGPVHHLL. Moreover, HLA-C*06:02-tetramers refolded with the SRGPVHHLL peptide stained KIR2DS1(+) KIR2DL1(−) NK cell clones, while no binding to KIR2DS1(−) KIR2DL1(−) NK cell clones was observed (Fig. 5c). Of note, as previous studies demonstrated that the co-expression of the inhibitor receptors KIR2DL2/KIR2DL3, KIR3DL1 and NKG2A can affect the outcome of KIR2DS1(+) NK cell degranulation 29 , we phenotyped all NK cell clones for these receptors. The presence or absence of KIR2DL2/KIR2DL3, KIR3DL1 and NKG2A on these clones did not affect the results (data not shown). All together, these data demonstrate that the HLA-C*06:02 peptide SRGPVHHLL enabled KIR2DS1-binding and resulted in the activation of KIR2DS1(+), but not KIR2DS1(−) NK cell clones derived from the same individual.

Discussion
The factors that determine the engagement of the activating NK cell receptor KIR2DS1 remain incompletely understood despite a growing number of genetic studies showing associations between KIR2DS1 and the outcome of various human diseases. In this study, we investigated the HLA class I molecules and HLA class I-presented peptides enabling KIR2DS1-binding and their influence on KIR2DS1(+) NK cell function. We demonstrate that KIR2DS1-binding is narrowly restricted to HLA-C2 molecules whereas KIR2DL1 showed a broader binding specificity for HLA-C2 but also for some HLA-C1 and HLA-B molecules. HLA-C2-presented peptides modulated both KIR2DS1-and KIR2DL1-binding to HLA-C2, and we identified one HLA-C*06:02-presented peptide (SRGPVHHLL) that strongly engaged KIR2DS1 and also KIR2DL1 binding. This synthetic peptide, predicted to bind HLA-C*06:02, did not correspond to any known human or viral sequence, and strongly activated primary KIR2DS1(+) NK cell clones in a peptide sequence-specific manner. Taken together, these data demonstrate peptide-dependent activation of the activating NK cell receptor KIR2DS1 and its inhibitory counterpart KIR2DL1.
A broad HLA class I-binding screen performed in this study demonstrated that KIR2DL1 and KIR2DS1 exhibited very similar binding specificities to HLA-C2 molecules, consistent with previous studies 16,17,22 . However KIR2DL1 showed a higher binding affinity for these HLA-C2 ligands, and a broader HLA class I binding specificity than KIR2DS1, as it also bound weakly to eight HLA-C1 complexes (HLA-C*01:02; *03:02; *03:03; *03:04; *07:02; *08:01; *12:03; *16:01) as well as two HLA-B complexes (HLA-B*46:01 and HLA-B*73:01). Given the opposite functions of the activating KIR2DS1 receptor and the inhibitory KIR2DL1 receptor, a very restricted set of ligands for the activating NK cell receptor KIR2DS1 might limit the risks of auto-immune reaction. Furthermore, both KIR2DS1 and KIR2DL1 shared the same pattern of peptide-specific recognition, which is consistent with previous studies 17,22,23,48 as they bound to the same HLA-C*06:02-presented peptide SRGPVHHLL. The functional assays performed using KIR2DS1ζ + and KIR2DL1ζ + Jurkat reporter cells also showed that KIR2DL1 had a stronger binding affinity than KIR2DS1, consistent with previous studies showing that inhibitory receptors have a stronger binding affinity than their corresponding activating NK cell receptor 17,30 for the same ligand. Of note, KIR2DL1 and KIR2DS1 are not always present at the surface of the same NK cell as KIR receptors are expressed stochastically, and it has been shown that around 10% of circulating NK cells expressed KIR2DS1 in absence of KIR2DL1 31 . Our data suggest that this subset of KIR2DS1(+) KIR2DL1(−) NK cells is able to mediate an effective effector function in response to their ligands. Notably, the tetramer staining performed using HLA-C*06:02 molecules refolded with the peptide SRGPVHHLL showed higher binding affinity for KIR2DS1 compared to KIR2DL1. This result might be linked to the observation that KIR2DS1 can assemble in larger clusters than KIR2DL1 at the surface of cells 32 , potentially favoring a stronger binding of HLA class I tetramers. Overall, our data showed that the NK cell receptors KIR2DS1 and KIR2DL1 shared the same ligand specificity for HLA-C2-presented peptides, but differed in their binding affinity.
Several studies using crystal structures of KIR2DL1 21 , KIR2DL2 19 , KIR2DL3 20 and KIR2DS2 25 have described that the KIR binding interactions with HLA class I presented epitopes are centered to the COOH-terminal end of the peptide which corresponds to peptide residues P7-P8 33 . In contrast, the specific recognition of HLA-presented peptide by TCR depends on the sequence of the entire peptide and is centered at the P4-P6 positions 34 . Our results showed that peptide residue P7 of the HLA-C*06:02 presented peptide SRGPVHHLL can modulate KIR2DS1 and KIR2DL1 binding, as previously described 22 , but that amino acid changes in residues P2 and P9 can also impact the binding of KIR2DS1 and KIR2DL1. The residues P2 and P9 are defined as anchor residues for HLA-C*06:02 binding [26][27][28] . Modifications of these residues may induce conformational changes of the peptide SRGPVHHLL presented by HLA-C*06:02, which can abrogate the binding to KIR2DS1 and KIR2DL1. The resolution of the crystal structure of KIR2DS1 in conjunction with HLA-C molecules presenting specific peptides will help clarifying these interactions. In summary, our results suggest that the HLA-C*06:02-presented peptide SRGPVHHLL can modulate KIR2DS1 and KIR2DL1 in a peptide sequence-specific manner.
Very few studies have investigated the ability of HLA class I-presented peptides to facilitate the binding of activating KIRs. Here we demonstrate that KIR2DS1-activation can be modulated by HLA-C2-presented peptides; however, despite screening over 500 virus-derived peptides, we did not identify any viral peptides binding to KIR2DS1. The peptide SRGPVHHLL identified to enable strong KIR2DS1-binding to HLA-C*06:02 was derived from a synthetic peptide library predicted to bind to HLA-C*06:02 26 , and does not match any known viral or human epitope. Several HIV-1-derived peptides presented by their respective HLA class I molecules have been demonstrated to modulate the binding of inhibitory KIRs, including KIR2DL2 35, 36 , KIR2DL3 37 and KIR3DL1 38,39 . Thus, it is remarkable that none of these peptides was able to engage the binding of KIR2DS1. Studies have indicated that viruses are able to select for sequence variants that enhanced the binding of inhibitory KIRs and thus inhibit the effector function of KIR+ NK cells, suggesting that viruses are able to escape NK cell-mediated immune pressure [38][39][40] . It might therefore be possible that HIV-1 has eliminated sequences that do allow binding to the activating KIR2DS1 receptor when presented by the HLA-C*06:02 molecule. To our knowledge, the only viral peptide presented by HLA class I and binding to an activating KIR that has been identified is a vaccinia virus-derived peptide in complex with HLA-A*11 binding to KIR2DS2 25 . Additionally, direct binding of primary activating KIR2DS1+ NK cells to virally infected cells has only been shown for EBV-transformed 221. HLA-C*04:01 17 , but no specific viral peptide was identified. Taken together, viruses might thus have evolved specific mechanisms to avoid recognition of viral peptides by activating NK cell receptors.
The role of activating KIRs has been extensively studied in the context of allogeneic stem cell transplantation to treat leukemia, especially in the setting of KIR ligands-mismatched donor/recipient pairs [41][42][43][44] . In particular, donor-derived KIR2DS1+ NK cells were shown to efficiently kill HLA-C2+ leukemia blasts, which indicate that activating KIRs can have a role in mediating anti-leukemic or anti-cancer effects 29 . The HLA-C*06:02 presented peptide SRGPVHHLL identified in this study mediated strong activation of primary NK cell clones, suggesting that this peptide could be potentially used to label specific tumors in order to enhance anti-tumor cytotoxic Scientific RepoRts | 7: 2414 | DOI:10.1038/s41598-017-02449-x activity by KIR2DS1(+) NK cells. The use of the synthetic SRGPVHHLL peptide as an enhancer for KIR2DS1(+) NK cell-mediated cytotoxicity might therefore provide a new perspective for NK cell immunotherapy.

Antibodies.
KIR-Jurkat reporter cell assay. KIRζ chimeric constructs for the generation of the KIR2DL1ζ + Jurkat reporter cells consisted of the extracellular and transmembrane domains of KIR2DL1*001 linked to the cytoplasmic tail of CD3ζ. KIR2DS1ζ + Jurkat reporter cells contained the extracellular domain of KIR2DS1*002, the transmembrane domains of KIR2DL1*001 linked to the cytoplasmic tail of CD3ζ. A sequence coding for the Zs Green protein was added to the KIR2DS1 chimeric construct to allow the differentiation between the KIR2DL1 and KIR2DS1 reporter cell line by GFP signal. Ligand engagement of KIR2DL1 or KIR2DS1 resulted in an activating signal that triggered CD69 expression. Measurement of KIR binding was assessed as Median Fluorescence Intensity (MdFI) of CD69. Reporter cells line were cultured overnight in RP20 (RPMI supplemented with 20% FBS) at 2.5*10 5 cells/ml to reduce background activation. KIR2DS1ζ + Jurkats cells or KIR2DL1ζ + Jurkats cells were co-incubated with target cells pulsed with peptides at a ratio effector/target 1/10 for 3 hrs at 37 °C under 5% CO2 in RP20. Cells were washed and stained for anti-CD3-BUV737, live/dead-BV510, anti-KIR2DL1/S1/ L3/S3/L5/S5-PE (HPMA4) and anti-CD69-BV421 20 mn at room temperature in the dark. After fixation in 4% paraformaldehyde, the samples were analyzed by flow cytometry (BD LSR Fortessa). KIR2DS1ζ + or KIR2DL1ζ + -Jurkat cells were used as negative control and KIR2DS1ζ + or KIR2DL1ζ + -Jurkat cells pulsed with 10 μl of HPMA4-conjugated beads (Dynabeads M-450 Tosylactivated, Invitrogen, conjugated following the manufacturer's instructions) were used as positive control.
Tetramer staining. HLA-C*06:02-SRGPVHHLL-PE tetramer was provided by the NIH Tetramer Core Facility and used for staining of target cell lines. Briefly, 2*10 5 cells of each target cell lines were incubated on ice for 5 mn in 50 µL blocking buffer (sterile PBS+ 10% human serum+ 3% fetal bovine serum (FBS)) in a 96 well plate. Cells were washed and resuspended in 50 µL blocking buffer and the tetramer was added at a dilution of 1/100 for 60 mn on ice in the dark. The samples were washed with FACS buffer (PBS+ 3% FBS) and stained with the corresponding antibodies for 30 mn on ice on the dark. After two washing rounds with FACS buffer, samples were fixed in 4% paraformaldehyde and analyzed by flow cytometry (BD LSR Fortessa). Measurement of HLA-C*06:02-SRGPVHHLL-PE tetramer binding was assessed as MdFI of PE.
Data acquisition, analysis and statistics. Flow cytometry data were analyzed using FlowJo software version 10.0.6 (Tree Star), and statistical analysis was performed using GraphPad Prism 5 (GraphPad Software). Each experiment was repeated independently 3 times except where stated otherwise. When indicated, statistical tests were performed assuming a non-parametric population using Kruskal-Wallis Test followed by post Dunn test analyzing all pairs of column. * , ** and *** corresponded to p < 0.05; p < 0.01 and p < 0.001, respectively.