Tissue-Specific Education of Decidual NK Cells

Abstract

During human pregnancy, fetal trophoblast cells invade the decidua and remodel maternal spiral arteries to establish adequate nutrition during gestation. Tissue NK cells in the decidua (dNK) express inhibitory NK receptors (iNKR) that recognize allogeneic HLA-C molecules on trophoblast. Where this results in excessive dNK inhibition, the risk of pre-eclampsia or growth restriction is increased. However, the role of maternal, self-HLA-C in regulating dNK responsiveness is unknown. We investigated how the expression and function of five iNKR in dNK is influenced by maternal HLA-C. In dNK isolated from women who have HLA-C alleles that carry a C2 epitope, there is decreased expression frequency of the cognate receptor, KIR2DL1. In contrast, women with HLA-C alleles bearing a C1 epitope have increased frequency of the corresponding receptor, KIR2DL3. Maternal HLA-C had no significant effect on KIR2DL1 or KIR2DL3 in peripheral blood NK cells (pbNK). This resulted in a very different KIR repertoire for dNK capable of binding C1 or C2 epitopes compared with pbNK. We also show that, although maternal KIR2DL1 binding to C2 epitope educates dNK cells to acquire functional competence, the effects of other iNKR on dNK responsiveness are quite different from those in pbNK. This provides a basis for understanding how dNK responses to allogeneic trophoblast affect the outcome of pregnancy. Our findings suggest that the mechanisms that determine the repertoire of iNKR and the effect of self-MHC on NK education may differ in tissue NK cells compared with pbNK.

FIGURE 1.

dNK show selective upregulation of specific iNKR compared with pbNK. (A) Freshly isolated lymphocytes from decidua or peripheral blood from the same donor were gated on live CD56⁺CD3⁻ cells and examined for expression of KIR2DL1, KIR2DL3, KIR3DL1, LILRB1, and NKG2A. (B) Overall frequency of CD56⁺CD3⁻ NK cells positive for the selected iNKRs in samples from blood (○) and decidua (●), expressed as a percentage of total NK cells (n = 53 and 61 for pbNK and dNK, respectively). (C) Effect of maternal HLA-C1 or -C2 alleles on overall frequency of expression of KIR2DL1 and KIR2DL3 was determined by stratifying donors from (B) according to the maternal HLA-C type. C1/C1 indicates donor had two C1 alleles; C2/X indicates either C2/C1 or C2/C2 genotype. Frequency of 2DL3⁺ NK cells was determined in pbNK (n = 24) and dNK (n = 34) by staining with mAb GL183 in patients who were genotyped as KIR2DL2⁻KIR2DS2⁻. C1/X indicates donor genotype was either C1/C1 or C1/C2. Horizontal bars indicate the mean percentage, and symbols indicate individual samples. *p < 0.05, **p < 0.001, Mann–Whitney U test.

FIGURE 2.

Expansion of specific iNKR subsets in decidual NK cells compared with matched pbNK. The expression of KIR2DL1, KIR2DL3/L2/S2, KIR3DL1, LILRB1, and NKG2A was examined by nine-color flow cytometry in matched pbNK and dNK from 21 donors. (A) Thirty-two distinct NK subsets were distinguished on the basis of the five iNKR in paired samples of pbNK and dNK (black and red graphs, respectively). The frequency of each subset from one representative donor is shown as a percentage of total NK cells. This provides a “fingerprint” of the iNKR repertoire, which is characteristic for each donor. The receptor combination for each subset is denoted by black-filled circles, with total number of iNKR in each subset indicated beneath. All NKG2A⁺ subsets are grouped to the right. The donor shown was genotyped as C1/C1. (B) The proportion of pbNK and dNK that express specific receptor combinations is shown as a percentage of total NK cells (mean ± SD, n = 21 matched pbNK and dNK pairs). The proportion of dNK that express two or more iNKR is significantly higher than in pbNK. (C) The percentage of dNK cells that are positive for Ki-67 was determined in each iNKR subset by intracellular staining. Control staining used isotype-matched mAb. The NK cells shown were gated on total KIR2DL1⁺ dNK as shown in Fig. 1A. (D) The percentages of dNK that are positive by intracellular staining for Ki-67 (●) and for CD122 (▴) are shown for all NKG2A⁺ dNK subsets (mean for each subset indicated by ● or ▴ ± SD, derived from n = 11 donors). (E) Mean fluorescence intensity (MFI) of staining of each subset is shown following staining for CD122 and c-Myc in the NKG2A⁺ dNK subsets from (D). For each donor, MFI of each subset was expressed relative to the subset expressing NKG2A alone, which was set at 100%. Relative MFI of CD122 and c-Myc is highly correlated (Spearman rank correlation coefficient, R² = 0.84, p < 0.001).

FIGURE 3.

dNK are educated differently by iNKR compared with pbNK. (A) Degranulation was measured by CD107a staining of 32 separate NK subsets from pbNK (○) and dNK (●), following coculture with K562 cells (n = 28 and 35, respectively). Only KIR2DL1⁺ subsets are shown (see Supplemental Fig. 3 for all 32 subsets). Subsets were defined by expression of five selected iNKRs as described in Fig. 1A. Frequency of CD107a⁺ cells is the percentage of NK cells in that subset that stain positive for CD107a (mean ± SD). Donors were not stratified according to HLA-C genotype. (B) Effect of donor’s HLA-C genotype on degranulation in pbNK and dNK in response to K562 was compared between donors stratified according to the maternal HLA-C genotype. C1/C1 indicates donor homozygous for C1 group; C2/X indicates C1/C2 or C2/C2. Subsets shown are KIR2DL1sp, KIR2DL1⁺NKG2A⁺, and KIR2DL1⁺LILRB1⁺. **p < 0.01, Mann–Whitney U test. Bar shows mean. (C) Effect of multiple iNKR on degranulation in response to K562 in pbNK (n = 12) and dNK (n = 14) from C1/C2 donors where both KIR2DL1 and KIR2DL3/L2 will have a maternal HLA-C ligand (●, receptor present on subset). There was a statistically significant difference in degranulation with increasing iNKR on each subset. **p < 0.001 for both pbNK and dNK, Friedman nonparametric ANOVA.

FIGURE 4.

NKG2A and LILRB1 have different effects on responses of pbNK and dNK subsets. (A) Effect of presence or absence of NKG2A on degranulation was measured by CD107a staining of 32 separate NK subsets in pbNK and dNK following coculture with K562 cells (n = 12 and 14, respectively; all donors genotyped as C1/C2). Subsets were defined as described in Fig. 1A. Average frequency of CD107⁺ cells is the percentage of the total NK cells in that subset that stain positive for CD107a (mean ± SD). The response of each subset lacking NKG2A expression (●) is compared with the corresponding subset expressing NKG2A (○). (B) The same analysis was performed to compare degranulation of subsets with (○) and without (●) LILRB1 in both pbNK and dNK.