KIR2DL2/2DL3-E35 alleles are functionally stronger than -Q35 alleles

KIR2DL2 and KIR2DL3 segregate as alleles of a single locus in the centromeric motif of the killer cell immunoglobulin-like receptor (KIR) gene family. Although KIR2DL2/L3 polymorphism is known to be associated with many human diseases and is an important factor for donor selection in allogeneic hematopoietic stem cell transplantation, the molecular determinant of functional diversity among various alleles is unclear. In this study we found that KIR2DL2/L3 with glutamic acid at position 35 (E35) are functionally stronger than those with glutamine at the same position (Q35). Cytotoxicity assay showed that NK cells from HLA-C1 positive donors with KIR2DL2/L3-E35 could kill more target cells lacking their ligands than NK cells with the weaker -Q35 alleles, indicating better licensing of KIR2DL2/L3+ NK cells with the stronger alleles. Molecular modeling analysis reveals that the glutamic acid, which is negatively charged, interacts with positively charged histidine located at position 55, thereby stabilizing KIR2DL2/L3 dimer and reducing entropy loss when KIR2DL2/3 binds to HLA-C ligand. The results of this study will be important for future studies of KIR2DL2/L3-associated diseases as well as for donor selection in allogeneic stem cell transplantation.

Besides KIR2DL2/L3, NK cells may express many other activating and inhibitory receptors that can also influence their functions. In this regard, Schonberg et al. reported that the presence of KIR2DL2 causes a major reduction in the frequency of NK cells expressing KIR2DL1 independent of the presence or absence of the C1 or C2 epitope 10 . Therefore, we analyzed the expression of various NK cell receptors in donors with different allelic groups of KIR2DL2/L3. Although there was natural variation in the expression of different NK receptors among different donors, we did not observe any statistically significant association with different allelic groups of KIR2DL2/L3 (Fig. 2C). Thus, the functional difference between E 35 and Q 35

In vivo anti-tumor activity of YT-Indy cells expressing different KIR2DL2/L3 alleles and mutants.
A mouse model was used to assess the anti-tumor activity of YT-Indy cells expressing different KIR2DL2/L3 alleles against 721.221 target cells expressing HLA-Cw07 ligand and luciferase. On day one, 8-to 12-week-old NSG mice were injected intraperitoneally with 5 × 10 5 target cells. Next day, YT-Indy cells expressing either KIR2DL3*001 (Q 35 ), KIR2DL3*005 (E 35 ), KIR2DL2*001 (E 35 ) or KIR2DL2*001-E35Q mutant were injected at a 10:1 E/T ratio. After 2 weeks, the bioluminescence signals of the mice that were injected with target cells only showed the highest signal whereas the group injected with YT-Indy cells containing empty vector (mock) showed the lowest signal (Fig. 4A). KIR2DL3*001 (Q 35 ) showed less inhibition than KIR2DL3*005 (E 35 ). While KIR2DL2*001 (E 35 ) showed strong inhibition (Fig. 4A,B), mutation of E 35 to Q 35 in KIR2DL2*001 markedly reduced inhibition to level similar to those of KIR2DL3*001 (Q 35 ), suggesting that E 35 is essential for durable inhibition in vivo.
Molecular modeling analyses suggest glutamic acid at position 35 of KIR2DL2/L3 forms hydrogen bond with histidine at position 55. To understand the structural mechanism, we performed molecular dynamic simulations on KIR2DL2*001-E 35 and KIR2DL2*001-Q 35 dimers. Overall, KIR2DL2*001-Q 35 showed substantially higher flexibility compared to KIR2DL2*001-E 35 (Fig. 5A,B). Amino acid position 35 of KIR2DL2/L3 is located in the loop region and does not directly interact with either HLA binding or dimerization interface. However, KIR2DL2/L3 alleles with E 35 form strong electrostatic interactions with the side chain of histidine at position 55 (Fig. 5C), which significantly stabilized not only the loop regions but also the HLA binding interface. Destabilization of this region greatly increases the flexibility of both loop regions and HLA binding interface (Fig. 5D), which reduces the binding affinity of KIR2DL2/L3 with HLA by increasing entropy loss during binding. Since it has been shown earlier that KIR2DL2/L3 dimer stability significantly influences its binding affinity with its ligands, the simulation data supports the hypothesis that the observed functional difference between KIR2DL2/L3-E 35 and -Q 35 arises from the protein stability change caused by E35Q substitution.

Discussion
KIR genes are highly polymorphic in nature exhibiting haplotypic and allelic variations 8,18 . The allelic diversity in each KIR gene ranges from 7 in KIR2DS3 to 76 in KIR3DL1 18 . Twelve KIR2DL2 alleles and 24 KIR2DL3 alleles have been identified thus far. Allelic variation may lead to functional diversity. For example a study by Moesta et al. reported that positions 16 and 148 accounted for KIR2DL2*001 being a stronger receptor for HLA-C ligand than KIR2DL3*001 9 . Schönberg K et al. showed that KIR2DL2 alleles might cross-react with HLA-C2 ligands but KIR2DL3 alleles could not 10 . A recent study by Fraizer et al. investigated the binding of 6 common KIR2DL2/L3 alleles with their HLA-C ligands 14 . They found that KIR2DL3*005 (which has proline and arginine at position 16 and 148 similar to those of KIR2DL3*001 but has glutamic acid at position 35 rather than glutamine) has affinity and avidity to HLA-C ligand as strong as KIR2DL2 alleles 14 . In line with these ligand-binding data, our functional cytotoxicity studies both in vitro and in vivo showed that KIR2DL2/L3 alleles with E 35 are functionally stronger than Q 35 alleles. NK cells licensed through stronger KIR2DL2/L3 alleles have higher cytotoxicity than those with weaker alleles against tumor cells with missing ligands, a finding in line with our previous observation that KIR2DL1 + NK cells licensed through stronger KIR2DL1-R 245 have higher cytotoxicity than KIR2DL1-C 245 19,20 .
KIR dimers are known to have higher affinity for HLA-C ligands than monomer. In the orthorhombic crystals of KIR2DL2, Snyder et al. found that two receptor molecules with the same orientation dimerize in such a way that the amino-terminal D1 domain of one receptor packs against the carboxyl-terminal D2 domain of the other molecule with their strands in an approximate orthogonal orientation, creating a D1/D2 heterodimer 21 . In another study, Fan et al. showed that covalently-linked KIR2DL1 dimer has higher affinity or avidity for HLA-Cw4 than monomer 22 . In the present study, molecular modeling analysis suggests that KIR2DL2/L3 alleles with the negatively charged E 35 form electrostatic interactions with the positively charged H 55 in KIR2DL2/L3 and Previous studies have shown that differences in KIR gene content are associated with the risk of many human diseases, including autoimmune diseases, inflammatory disorders, infectious diseases, immunodeficiency, cancer, and reproductive disorders 23 . KIR2DL2/L3 is specifically known to be associated with Crohn's disease 24 , multiple sclerosis 25 , chronic myeloid leukemia 26 , primary Sjogren's syndrome 27 , lupus 28 , birdshot chorioretinopathy 29 , rheumatoid arthritis 30 , cervical intraepithelial neoplasia 31 , scleroderma 32 , and psoriasis 33 . KIR2DL2/L3 is also reported to be associated with herpes simplex virus 1 34 as well as human immunodeficiency virus 1 35 infections. Moreover, KIR2DL2/L3 are important determinant for donor selection in hematopoietic stem cell transplantation (HSCT). In this regard, we have previously showed that a single amino acid can be used to divide the KIR2DL1 alleles into functionally strong and weak groups 11 . In the allogeneic HSCT setting, this biologic difference can be translated to markedly different patient survival and risk of leukemia relapse 20 . Taken together, our findings herein will have significant implications in donor selection and outcome evaluation in HSCT and NK cell therapy (KIR2DL1-R 245 and KIR2DL2/L3-E 35 are predicted to be favorable). The results will also be useful in setting the framework for future NK cell research in various clinical settings, including susceptibility and outcome of cancers, pregnancy complications, chronic viral infections, and autoimmune diseases.

Materials and Methods
Statement. Animal experiment was carried out according to the protocols that were approved by The Institutional Animal Care and Use Committee of St. Jude Children's Research Hospital.

DNA constructs. Peripheral blood mononuclear cells were obtained from healthy human donors with
informed consent under a protocol approved by the institutional review board at St Jude Children's Research Hospital, in accordance with the Declaration of Helsinki. Total RNA was extracted from cells expressing KIR2DL2/L3 using RNA extraction kits (QIAGEN). cDNA of various alleles of KIR2DL2/L3 was generated from the RNA and cloned into mammalian expression vector pcDNA3 (Invitrogen). The identities of the KIR2DL2/L3 alleles were confirmed by sequencing. Specific amino acids were substituted into KIR2DL2/L3 using recombinant polymerase chain reaction. HLA-Cw7 cDNA was amplified from a normal human peripheral blood cDNA pool as described for KIR2DL2/L3 alleles, confirmed by sequencing, and cloned into retroviral vector MSCV-IRES-GFP (received from vector laboratory, St Jude Children's Research Hospital). University) were cultured in RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 1 mM penicillin/streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate, and 1% minimum essential medium nonessential amino acids (Invitrogen). B-lymphoblastic cell line 721.221 was purchased from the International Histocompatibility Working Group and cultured in RPMI 1640 supplemented with 20% FBS and 1 mM penicillin/streptomycin. YT-Indy cells were transfected with pcDNA3 vector containing various KIR2DL2/L3 alleles by electroporation (Gene Pulser II; Bio-Rad). Stable cell lines were generated by selection in Geneticin (Invitrogen). The 721.221 cells were transduced with retroviral vector MSCV-IRES-GFP containing HLA-Cw7. High-expressing cells was sorted by flow cytometric cell sorting using the monoclonal antibody (mAb) against HLA-C (One Lambda).

KIR2DL2/L3 single nucleotide polymorphism (SNP) and genotyping assay.
To detect the presence of various alleles of KIR2DL2/L3 with glutamic acid or glutamine at position 35, a single-nucleotide mismatch detection assay was developed as described previously 19,36 . Briefly, primers for the assay were designed in such a way that they amplified all the alleles of the KIR2DL2/L3 gene as well as the amplicon containing the polymorphic region of interest. The forward primer was 5′ -CATCCTGCAATGTTGGTCAG-3′ and the reverse primer was 5′ -CAAGGTCTTGCATCATGGGA-3′ . The probe for KIR2DL2/L3 alleles with a glutamic acid at position 35 was 6Fam-CAGGTTTGAGCACT-MGBNFQ and for those with a glutamine at the same position was VIC-CAGGTTTCAGCACT-MGBNFQ. Each assay reaction mix contained a 250 nM probe concentration and 100 ng of genomic DNA in 1× TaqMan genotyping master mix (Applied Biosystems). The assay was performed on an HT7900 Sequence Detection System (Applied Biosystems) following the allelic discrimination assay protocol provided by the manufacturer. KIR2DL2/L3 genotyping was performed by a PCR-SSP method as described previously 17 .  anti-CD14 (MphiP9). Flow cytometry analyses were conducted with LSRII (BD Biosciences) and FlowJo 8.8.6 software (Tree Star).

Cytotoxicity assay.
To determine the functional differences amongst different KIR2DL2/L3 and mutated alleles, cytotoxic activity was measured by using the DELFIA BATDA reagent (PerkinElmer Life and Analytical Sciences) following the manufacturer's instruction. BATDA-labeled B-lymphoblastic cell lines 721.221 with or without HLA-C ligands were used as target cells at an effector : target (E:T) ratio of 20:1 for 2 hours at 37 °C. The fluorescence signals were measured using a Wallac Victor 2 Counter Plate Reader (PerkinElmer Life and Analytical Sciences). CD107 degranulation assay. NK cells were tested for their cytolytic potential with the CD107 degranulation assay. Effector cells were cocultured with target cells at 1:1 ratio in the presence of anti-CD107-FITC or -APC antibodies. After 1 hour of co-culture, GolgiStop (BD Biosciences) was added and the cells were incubated for 4 more hours. The cells were then harvested, stained, and analyzed for CD107.
In vivo experiments. Non-obese diabetic/severe combined immunodeficient (NOD/SCID) IL-2 γc−/− mice, 8-to 12-weeks old, were used as animal model. 721.221-Cw7-luciferase cells were used as target cells. Mice were γ -irradiated at a dose of 200 cGy 1 day prior to intraperitoneall injections of 5 × 10 5 721.221-Cw7 cells. On the following day, 5 × 10 6 YT-Indy cells expressing different KIR2DL2/L3 alleles (effector cells) were injected intraperitoneally (E:T = 10:1). Disease progression in the injected mice was monitored by bioluminescence imaging (Xenogen, PerkinElmer Life and Analytical Sciences). The mice were sacrificed when they displayed signs of significant tumor progression. The experiments were terminated after 3 weeks as bioluminescence reached saturation point.

Molecular modeling.
The structural effects of E35Q mutation of KIR2DL2 were simulated by molecular dynamics simulation 37 . The dimer structure of both E35 and Q35 were derived from crystal structure of KIR2DL2-HLA complex (PDB ID: 1EFX) 21,38 . The TIP3P water solvent box with 10 angstroms from the protein were added to each system, and counterions were used for neutralizing the charges. Each system were thoroughly minimized and heated to 300 K, and simulated for 25 ns by Langevin dynamics. The backbones root mean squire fluctuations (RMSFs) of CA atoms were calculated to evaluate protein stabilities.
Statistical analysis. Statistical significance between 2 groups was calculated using Student T test. The nominal significance level was set at 0.05.