NK cell tolerance of self-specific activating receptor KIR2DS1 in individuals with cognate HLA-C2 ligand

NK cell donors

NK cells were obtained from 7 individuals (5 healthy donors and 2 transplant recipients). HLA class I genotyping was performed on genomic DNA by a combination of PCR amplification with sequence-specific primers or sequence-specific oligonucleotide probes (17). KIR genotyping was performed by KIR sequence-specific primers (KIR genotyping SSP Kit, Invitrogen) and KIR haplotypes and genotypes were assigned (18) (Table I). NK cells from healthy donors were negatively selected from freshly isolated PBMC obtained from 30 ml peripheral blood, using a cocktail of magnetically labeled mAbs specific for non-NK lineage antigens (Miltenyi Biotec) (19). For all experiments, post-isolation NK cell purity was >90%. NK cells from transplant recipients were directly FACS-sorted from bulk PBMC (see NK cloning).

BaF/3 IL-15Rα/IL-15 transfectants

pSFG retroviral vectors containing full length cDNA of human IL-15Rα or IL-15 (kindly provided by Dr. Thomas A. Waldmann, Metabolism Branch, National Cancer Institute) were transfected into Phoenix E packaging cell line, to produce retroviral supernatants. BaF/3 cells were incubated with retroviral supernatants, for 6–8 h, in fibronectin-coated plates (Takara Biomedicals). Clones of IL-15Rα/IL-15 double-transfected pre-B-lymphocyte BaF/3 cells (BaF/3 IL-15Rα/IL-15) were obtained by limiting dilution, and stable expression of IL-15 and IL-15Rα was confirmed by monthly mAb staining. The cell line was maintained in RPMI 1640 supplemented with 10% FCS, 100 U/ml penicillin, 0.1 mg/ml streptomycin and 2 mM L-glutamine (all provided by the Core Media Preparation Facility, Memorial Sloan-Kettering Cancer Center).

NK cloning

NK clones were developed following single cell deposition (Fig. 1) and propagated by IL-15 trans-presentation. NK cell subpopulations displaying specific combinations of KIR/NKG2A expression were identified by the following mAbs: CD3-PE/TexasRed (S4.1, Invitrogen), CD56-PE/Cyanine7 (MEM-188, BioLegend), 2DL1/S1-PerCP/Cyanine5.5 (HP-MA4, BioLegend), 2DL1-Allophycocyanin (143211, R&D Systems), 2DL2–3/S2-FITC (CH-L, Miltenyi Biotec), 3DL1-Alexa Fluor700 (DX9, BioLegend), 3DL1/S1-PE (Z27, Beckman Coulter), NKG2A-PE (131411, R&D Systems), CD85j/ILT2 (LILRB1)-PE (HP-F1, Beckman Coulter). Single NK cells from selected subpopulations were FACS-sorted (Aria III, BD Biosciences) and deposited into U-shaped polystyrene 96-well plates (one cell/well) containing 100 μl CellGro SCGM medium (CellGenix GmbH) supplemented with 10% heat-inactivated AB human serum (Gemini Bioproducts), 100 U/ml penicillin, 0.1 mg/ml streptomycin and 2 mM L-glutamine. The following feeders were added to the medium: 10⁴ allogeneic EBV-B lymphoblastoid cell line (EBV-BLCL) (JY); 4 × 10⁴ PBMC obtained from 3 different donors, and 3 × 10³ BaF/3 IL-15Rα/IL-15 cells. Feeders were gamma irradiated (EBV-BLCL and PBMC: 5.2 Gray; BaF/3 IL-15Rα/IL-15: 13.9 Gray). After sorting, plates were centrifuged at 500 rpm for 1 min and incubated in a 37 °C, 5% CO₂ humidified atmosphere. On d 5, 100 μl fresh medium and irradiated EBV-BLCL (10⁴), PBMC (4 × 10⁴) and BaF/3 IL-15Rα/IL-15 (3 × 10⁴) were added to each well. On d 10, 80 μl supernatant were removed and substituted with medium and irradiated feeders, as on d 5. On d 15, NK cell growth could be detected by a colorimetric change (purple to yellow) of the microculture supernatant. Proliferating NK clones were collected, transferred in 48- well plates and supplemented with 800 μl medium and irradiated BaF/3 IL-15Rα/IL-15 (5 × 10⁵). On d 22, 300 μl supernatant were removed from all wells, and replaced with 500 μl medium and irradiated BaF/3 IL-15Rα/IL-15 (5 × 10⁵). Between d 28 and d 32, NK cells were screened by flow cytometry to determine viability, clonality and receptor expression. NK clones were harvested, functionally characterized and cryopreserved for subsequent molecular studies.

Characterization of NK clones

Quantitative PCR

IL-15 trans-presentation supports generation of 2DS1^pos NK clones

2DS1^pos clones have previously been obtained from donors lacking cognate HLA-C2 ligand (i.e., donors homozygous for the HLA-C1 ligand). In contrast, very few 2DS1^pos clones were obtained from donors expressing HLA-C2 (19). Since IL-15 trans-presentation is the major growth and survival signal for NK cells (24–27), we investigated, if human NK cloning efficiency and clone survival could be enhanced by IL-15 trans-presentation in vitro. Trans-presentation was achieved by co-culture of FACS-sorted NK cells with murine BaF/3 cells transfected with human IL-15Rα and human IL-15. This procedure supported clone development from all donors, irrespective of their HLA-C genotype. Each clone reached 0.25–4 × 10⁶ cells, and the overall cloning efficiency was 35%–40%. Clones were developed from seven donors, representing the three HLA-C genotypes: C1:C1; C1:C2 and C2:C2. HLA-KIR ligand groups and KIR genes for each NK donor are listed in Table I. We analyzed 386 clones, which included the 2DS1^SP phenotype and other 2DS1^pos phenotypes (Table II). For example, 2DS1^SP clones were obtained from donors with each HLA-C genotype (Table II, Columns A and B). Similarly, twenty-four 2DS1^SP, which also expressed the inhibitory receptor CD94/NKG2A were obtained (Table II, Column F). Accordingly, clones with a broad KIR repertoire, including 2DS1, can be obtained from donors with any HLA-C genotype, when IL-15 trans-presentation is the NK growth factor.

Frequency of 2DS1^SP NK clones with anti-HLA-C2 cytotoxicity is only decreased in donors homozygous for HLA-C2

A total of 91 2DS1^SP clones were isolated of which 56 had anti-HLA-C2 cytotoxicity (Table III). Clones with anti-HLA-C2 cytotoxicity were obtained from any donor, regardless of the HLA-C genotype. Anti-HLA-C2 cytotoxicity was detected in 29 of 42 clones with the C1:C1 genotype (69%) and in 19 of 22 clones with the C1:C2 genotype (86%). In contrast, anti-HLA-C2 cytotoxicity was only observed in 8 of 27 clones with the C2:C2 genotype (30%) (C1:C1 vs C2:C2, p=0.001 and C1:C2 vs C2:C2, p<0.0001) (Fig. 2A and Table IIIA).

The frequency of anti-HLA-C2 cytotoxicity among 2DS1^SP, C1:C2 clones was similar to the frequency observed among C1:C1 clones (Fig. 2A, Left and Center). This demonstrates that clonal deletion or clonal anergy is not characteristic for 2DS1^SP clones from donors heterozygous for HLA-C2. The 22 clones from C1:C2 heterozygous donors were tested on HLA-C2 heterozygous and homozygous target cells. The C1:C2 clones were significantly less frequently cytotoxic against target cells with the autologous C1:C2 genotype (p<0.0001) (Fig. 2A, Center and Fig. 2B; Table IIIB). Thus, 2DS1^SP clones with anti-HLA-C2 reactivity derived from donors with the C1:C2 genotype are rarely cytotoxic to autologous targets. This decrease in frequency of anti-HLA-C2 cytotoxicity cannot be ascribed to the effect of inhibitory KIR expressed by the clones, since they all are 2DS1^SP.

We finally determined the effect of inhibitory KIRs with ligand specificity for non-self-HLA class I on the function of 2DS1^pos clones. Thirteen 2DS1^pos, C2:C2 clones, which also expressed the inhibitory receptor 2DL3 with ligand specificity for HLA-C1, were obtained. Six of 13 (46%) clones had anti-HLA-C2 reactivity, which is not significantly different from the results obtained with 2DS1^SP, C2:C2 clones (Fig. 2A, Right and Fig. 2C; Table IIIC). Therefore, 2DS1^pos, HLA-C2 homozygous clones with non-self inhibitory KIR display comparable anti-HLA-C2 reactivity as 2DS1^SP clones from the same donor.

2DS1^SP NK clones with anti-HLA-C2 reactivity are present in recipients of 2DS1^pos allogeneic hematopoietic stem cell transplantation

Allogeneic, myeloablative hematopoietic stem cell transplantation (HCT) provides a possibility for evaluating de novo development of donor-derived 2DS1^pos NK cells in the presence of cognate HLA-C2 ligand. We investigated two cases, where the graft was obtained from 2DS1^pos C1:C2 donors. 2DS1^SP NK cells were identified post-HCT in both recipients (Fig. 2D). In the first case, ten 2DS1^SP clones were obtained 100 d post-HCT from a HLA-C2 homozygous recipient. Four of the 2DS1^SP clones displayed anti-HLA-C2 cytotoxicity (Fig. 2D, Case 1). In the second case, four 2DS1^SP clones were obtained from a HLA-C1:C2 heterozygous recipient, 200 d post-HCT. Two clones displayed anti-HLA-C2 cytotoxicity (Fig. 2D, Case 2). Therefore, donor-derived 2DS1^SP NK cells with ability to mediate anti-HLA-C2 cytotoxicity can be identified in an allogeneic host that expresses the cognate HLA-C2 ligand.

2DS1^pos NK clones expressing at least one inhibitory KIR for self-HLA class I are tolerant

Among the inhibitory KIR with HLA class I ligand specificity, only 2DL1 and 3DL1 can be individually recognized by mAbs. The remaining inhibitory KIR 2DL2, 2DL3 and the activating receptor 2DS2 cannot be distinguished by monospecific Abs. Similarly, 2DS1 is not distinguishable from 2DL1, when both receptors are present on the same cell. KIR phenotyping was therefore supplemented with determination of mRNA copy numbers for each of the KIRs with ambiguous phenotypes. Absolute RT-qPCR quantification assays were performed for 2DL1 (166 clones); 2DS1 (285 clones) and 2DL2-3 (229 clones) KIR transcripts. The mRNA copy numbers for such KIRs were determined and the minimal copy number associated with KIR surface expression was identified, as described in Materials and Methods and in Figure 3. These minimal copy number values were used as reference to assign a KIR receptor phenotype to a total of 29 clones, whose surface expression of one or more KIR could not be unambiguously determined by flow cytometric analysis alone. This method allowed distinction of 2DL3 expression from 2DS2 expression in 16 clones, and detection of 2DS1 expression in 12 2DL1^pos clones. Moreover, we could identify 2DL1 and 2DS1 expression in 3 clones, that were characterized by a mAb with specificity for both 2DL1 and 2DS1.

Forty-six 2DS1^pos clones expressing at least one inhibitory KIR for self-HLA class I were obtained. Thirty-two clones were heterozygous for HLA-C2 and 14 clones were homozygous for HLA-C2. Only one clone displayed anti-HLA-C2 cytotoxicity (2%) (Table IV). Therefore, 2DS1^pos NK cells expressing at least one inhibitory KIR for self-HLA class I very rarely mediate anti-HLA-C2 cytotoxicity, when the donor has HLA-C2. In contrast, 2DS1^pos, HLA-C1 homozygous clones co-expressing inhibitory KIR for self-HLA class I frequently display anti-HLA-C2 cytotoxicity (Table V and Ref. 19).

Anti-HLA-C2 cytotoxicity of 2DS1^SP clones expressing the inhibitory receptor LILRB1

The leukocyte Ig-like receptor (LILR) B1 is expressed on NK cell subsets and other cells belonging to the myeloid and lymphoid lineage (28, 29), and delivers inhibitory signals upon interaction with a wide range of HLA class I antigens (30, 31). Twenty-two 2DS1^SP clones expressed LILRB1. Eight clones were obtained from HLA-C1 homozygous donors and 14 clones from HLA-C2 homozygous donors. All clones from C1 donors displayed anti-HLA-C2 cytotoxicity. In contrast, anti-HLA-C2 cytotoxicity was only observed in 5 of 14 clones from C2 donors (C1:C1 vs C2:C2, p=0.003). Therefore, LILRB1 may in some instances contribute to inhibition and maintaining tolerance of 2DS1 signals in HLA-C2 homozygous donors.

Effect of inhibitory receptor CD94/NKG2A on tolerance development of HLA-C2 homozygous 2DS1^SP NK cells

The reduced frequency of anti-HLA-C2 reactivity observed in HLA-C2 homozygous, 2DS1^SP clones is consistent with NK tolerance observed in mice transgenic for activating receptor ligands (7–11). In humans, the inhibitory receptor CD94/NKG2A could potentially counteract 2DS1 activation by the HLA-C2 ligand. We first determined if the 2DS1 receptor is signaling-competent in 2DS1^SP, HLA-C2 homozygous clones expressing CD94/NKG2A. EB6 mAb cross-linking of the 2DS1 receptor in the presence of ICAM-I induces Ca⁺⁺ flux (Materials and Methods). This activation signal is inhibited when HLA-E, the ligand for CD94/NKG2A (32), is added (Fig. 4A). Similar results were obtained with 3 additional clones, demonstrating that the 2DS1 receptor is signaling-competent. Next, we determined the correlation between NKG2A expression levels, and anti-HLA-C2 reactivity of 2DS1^SP/NKG2A^pos clones. NKG2A mRNA transcript copy numbers correlated well with NKG2A receptor MFI in 2DS1^SP/NKG2A^pos clones from C1:C1 and C2:C2 donors (p=0.002) (Fig. 4B). This data validated the inclusion of NKG2A mRNA copies for our correlation studies. We expected that high cytotoxicity would be observed in clones with low NKG2A mRNA copies. However, NKG2A mRNA copy numbers were not found to correlate with anti-HLA-C2 cytotoxicity in 2DS1^SP, C2:C2 clones (Fig. 4C). Therefore, expression of CD94/NKG2A inhibitory receptor does not predict if 2DS1^SP, HLA-C2 homozygous clones mediate anti-HLA-C2 reactivity, or if they are tolerant to HLA-C2.

In order to directly test the inhibitory function of CD94/NKG2A on 2DS1^SP/NKG2A^pos, HLA-C2 homozygous clones, we determined the effect of anti-NKG2A F(ab′)2 fragment on cytotoxicity of 10 non-cytolytic (Fig. 4D) and 14 cytolytic (Fig. 4E) clones. Anti-HLA-C2 cytotoxicity was determined using HLA-A*02:01 homozygous target (BLCL 9036), which expresses several HLA class I alleles with HLA-E binding leader peptides (33). HLA-E expression on this target was confirmed by mAb staining. Only one non-cytolytic clone changed from non-cytolytic to cytolytic, when the CD94/NKG2A receptor was blocked with anti-NKG2A F(ab′)2 (Fig. 4D). Furthermore, all 14 2DS1^SP clones with anti-HLA-C2 reactivity displayed enhanced cytotoxicity following blocking with anti-NKG2A F(ab′)2 (Fig. 4E). Therefore, CD94/NKG2A only provides a modulatory, attenuating effect on 2DS1-mediated anti-HLA-C2 cytotoxicity, and is not a major factor controlling 2DS1 tolerance to HLA-C2 in HLA-C2 homozygous donors. These results also agree with a recent report of a single 2DS1^pos/NKG2A^pos clone, where 2DS1-mediated cytotoxicity was unaffected by presence of CD94/NKG2A (34).

2DS1 activation, missing self-HLA class I recognition and KIR inhibition

We investigated the relationship between 2DS1 activation by HLA-C2 ligand, NK activation by missing self-HLA class I and inhibition of NK cell activation by inhibitory KIR to self-HLA class I. Eight 2DS1^pos clones that also expressed the inhibitory receptor 3DL1 were obtained from a donor heterozygous for C1:C2 and homozygous for HLA-Bw4. The clones were tested against a panel of target cells homozygous for the 2DS1 activating ligand (i.e., C2:C2 homozygous) or lacking the activating ligand (i.e., C1:C1 homozygous). Presence of inhibitory ligand for 3DL1 (i.e., Bw4 homozygous) or absence of 3DL1 ligand (i.e., Bw6 homozygous) was similarly tested in different combinations of target cells (Fig. 5). Six of eight clones were inhibited by a target homozygous for HLA-C2;Bw4 (Fig. 5, Left), consistent with the finding that clones with inhibitory KIR for self-HLA class I in most instances do not respond to HLA-C2 activation (Table IV). The same clones were tested on a target homozygous for HLA-C2 but lacking the HLA-Bw4 ligand for 3DL1 (i.e., HLA genotype C2:C2;Bw6:Bw6). All eight clones displayed anti-HLA-C2 cytotoxicity, demonstrating the combined activating effect of 2DS1 signaling and recognition of “missing self-HLA class I” (Fig. 5, Center). Finally, the effect of lack of HLA-C2 ligand and absence of HLA-Bw4 on target cells (i.e., no stimulation of 2DS1 but recognition of “missing self-HLA class I”) is shown. Three clones were cytolytic, while 5 were inhibited (Fig. 5, Right). Therefore, “missing self-HLA class I” recognition in the absence of 2DS1 activation provides a variable activation signal.

Tolerance of 2DS1^SP clones to cognate ligand in HLA-C2 homozygous donors is not dependent upon ligand-mediated down-regulation of the 2DS1 receptor

Studies in transgenic mice have demonstrated ligand-mediated down-regulation of the activating NK receptor (10, 11). A similar mechanism could be involved in 2DS1 tolerance to HLA-C2. MFI of 2DS1 expression was determined on the eight C2:C2 2DS1^SP clones displaying anti-HLA-C2 cytotoxicity, and on the nineteen C2:C2 2DS1^SP clones lacking anti-HLA-C2 cytotoxicity to the HLA-C2 homozygous target. 2DS1 expression levels were similar in these HLA-C2 homozygous 2DS1^SP clones, irrespective of their anti-HLA-C2 responsiveness (Fig. 6).

3DS1 does not contribute to anti-HLA-C2 reactivity of 2DS1^SP clones

Because the genes encoding the two activating receptors 2DS1 and 3DS1 are in strong positive genetic linkage disequilibrium (35, 36), they frequently occur together. It is therefore potentially difficult to distinguish between a 3DS1 and a 2DS1-mediated effect. The minimal 3DS1 mRNA copy numbers needed for expression of 3DS1 receptor was 98 (Fig. 3 and Materials and Methods). This value was used as reference for determination of 3DS1 receptor expression. Ninety 2DS1^SP clones from donors with different HLA-C genotypes were tested for 3DS1 expression. Sixty-five clones (72%), had the 2DS1^SP/3DS1^pos phenotype with 3DS1 mRNA copy numbers consistent with 3DS1 cell surface expression (median: 544; range: 102–2340). The remaining twenty-five clones had the 2DS1^SP/3DS1^neg phenotype. Their 3DS1 mRNA copies were either undetectable, or present in numbers below the calculated minimal value, needed for expression (median: 0; range: 0–79). The frequency of anti-HLA-C2 cytotoxicity was compared between 2DS1^SP/3DS1^pos and 2DS1^SP/3DS1^neg clones. The analysis was done separately for different HLA-C groups of donors, because the HLA-C genotype affects the frequency of anti-HLA-C2 reactive 2DS1^SP clones (Fig. 2). It is demonstrated in Table VI, that presence or absence of 3DS1 does not affect anti-HLA-C2 reactivity in 2DS1^SPclones. This was observed in clones from HLA-C1 positive (i.e., C1:C1 and C1:C2), and clones from HLA-C2 homozygous donors. Therefore, 3DS1 does not contribute to anti-HLA-C2 cytotoxicity by 2DS1^SP clones.