Class I HLA haplotypes form two schools that educate NK cells in different ways

Abstract

Natural killer (NK) cells are lymphocytes having vital functions in innate and adaptive immunity, as well as placental reproduction. Controlling education and functional activity of human NK cells are various receptors that recognize HLA class I on the surface of tissue cells. Epitopes of polymorphic HLA-A,-B and -C are recognized by equally diverse killer cell immunoglobulin-like receptors (KIR). In addition, a peptide cleaved from the leader sequence of HLA-A,-B or -C must bind to HLA-E for it to become a ligand for the conserved CD94:NKG2A receptor. Methionine/threonine dimorphism at position -21 of the leader sequence divides HLA-B allotypes into a majority having -21T that do not supply HLA-E binding peptides and a minority having -21M, that do. Genetic analysis of human populations worldwide shows how haplotypes with -21M HLA-B rarely encode the KIR ligands: Bw4⁺HLA-B and C2⁺HLA-C KIR. Thus there are two fundamental forms of HLA haplotype: one preferentially supplying CD94:NKG2A ligands, the other preferentially supplying KIR ligands. -21 HLA-B dimorphism divides the human population into three groups: M/M, M/T and T/T. Mass cytometry and assays of immune function, shows how M/M and M/T individuals have CD94:NKG2A⁺ NK cells which are better educated, phenotypically more diverse and functionally more potent than those in T/T individuals. Fundamental new insights are given to genetic control of NK cell immunity and the evolution that has limited the number of NK cell receptor ligands encoded by an HLA haplotype. These finding suggest new ways to dissect the numerous clinical associations with HLA class I.

Figure 1. Alleles encoding HLA-B -21M and C2⁺HLA-C segregate on different haplotypes outside of Africa

A. The C1/C2 dimorphism at position 80 of HLA-C and the -21 M/T dimorphism of HLA-B define four HLA B-C haplotypes. Shown are the proportions of each haplotype in major human population groups: 51 populations and 16,384 HLA haplotypes were analyzed. B. plots the linkage disequilibrium (D’) between the HLA-B and HLA-C dimorphisms: Africans (dark green circles), Asians (orange circles), admixed US populations of African ancestry (light green circles) and Asian ancestry (yellow circles), all other populations (black circles). The haplotype frequencies used to calculate D’ are given in Figure S1. C. Described are characteristics of the five haplotypes that encode -21M HLA-B and C2⁺HLA-C. D. Frequencies of the three -21 HLA-B genotypes in major population groups: -21T/T homozygotes (red circles), -21M/T heterozygotes (blue circles) and-21M/M homozygotes (black circles). (8,192 genotypes were analyzed)

Figure 2. -21 HLA-B dimorphism modulates expression of HLA-E and CD94:NKG2A

Mass cytometry analysis of PBMC from 20 donor cohorts defined by -21 HLA-B genotype: homozygous -21T (T/T) red circles, heterozygous -21M/T (M/T) blue circles and homozygous -21M (M/M) black circles. A. Compares the cell-surface HLA-E expression given by the median signal intensity (msi). Donors are also defined by 107 HLA-E genotype: homozygous 107 R (R/R) circles, heterozygous 107 R/G (M/T) triangles and homozygous 107 G (G/G) squares. B. Compares the frequency of CD94:NKG2A⁺NK cells. C. Compares the proportion (%) of NK cells expressing CD94:NKG2A with their level of surface expression (msi). The thick black line shows the goodness of fit (R²) determined by Spearman correlation analysis. The thin red lines denote the 95% confidence interval (range 0.44–0.77). The correlation coefficient (R²) is shown with a two-tailed p value. All donors were CMV negative.

Figure 3. -21 HLA-B dimorphism correlates with differences in HLA-C expression

Surface expression of HLA-C by donor PBMC was determined by flow cytometry and quantified by median signal intensity (msi). A. Donors are grouped by -21M HLA-B genotype. B. Donors are grouped by presence and absence (x) of HLA-C*07. C. Donors are grouped according to the C1 and C2 epitopes of HLA-C. D. Donors grouped by genotype of the miRNA-148a binding site in the 3’-untranslated region of HLA-C: ’I’ is the intact functional allele, ‘D’ is the non-functional allele that contains a deletion. Statistical significance was determined by two-tailed unpaired Mann Whitney tests and given as mean p values ± standard error. E, F and G compares the surface expression of HLA-C, with HLA-C genotype for M/M donors, M/T donors and T/T donors, respectively. Every donor is represented by two red circles one for each HLA-C allele, and placed above the allele’s name. HLA-C homozygotes have two circles with identical msi in the same column.

Figure 4. -21 HLA-B dimorphism modulates the co-expression of inhibitory receptors by educated CD94:NKG2A⁺ NK cells

A. NK cell expression of CD4:NKG2A. B. Coexpression of KIR2DL1 and CD94:NKG2A. C. Coexpression of KIR3DL1 and CD94:NKG2A. D. Coexpression of KIR3DL2 and CD94:NKG2A. E. Coexpression of LILRB1 and CD94:NKG2A. In each panel there are three diagrams representing Spanning-tree Progression Analysis of Density normalized Events (SPADE) for CD94:NKG2A⁺NK cells. Each NK cell subpopulation is represented by a circle, its area corresponding to the subpopulation’s size. Shown are data for representative M/M (left), M/T (left center) and T/T donors (right center). On the right are histograms showing the phenotype diversity in the three donor cohorts. Phenotypic data was obtained by mass cytometric analysis using antibodies specific for 35 cell-surface markers.

Figure 5. -21M HLA-B drives expansion of educated CD94:NKG2A⁺ NK cells expressing enhanced levels of activating receptors

SPADE for three representative M/M (left), M/T (center left) and T/T (center right) donors combined with histograms giving the phenotypic diversity of CD94:NKG2A⁺ NK cells in the cohorts of 20 M/M, M/T and T/T donors (right). A. CD57-expressing CD94:NKG2A⁺ NK cells. B. CD16-expressing CD94:NKG2A⁺ NK cells. C. CD122 CD57-expressing CD94:NKG2A⁺ NK cells. D. Frequencies of CD57⁻ and CD57⁺ CD94:NKG2A⁺ NK cells. E. Frequencies of CD16⁻ and CD16⁺ CD94:NKG2A⁺ NK cells. F. Frequencies of 2B4⁻ and 2B4⁺ CD94:NKG2A⁺ NK cells. G. Frequencies of NKG2D⁻ and NKG2D⁺ CD94:NKG2A⁺ NK cells. Statistical significance (p values) derives derived from two-tailed unpaired Mann Whitney (A–C) and paired Wilcoxon (D–G) tests (mean ± SE).

Figure 6. -21 HLA-B dimorphism modulates NK cell effector function

A. PBMC were cultured for 6 hours in an ADCC assay with Raji cells (E:T = 10:1) coated with with anti-CD20 Ab or murine IgG (IgG) (range: 0.32–10µg/ml) and the CD3⁻CD56^dim NK cells assayed for surface expression of CD107a. B. PBMC were cultured for 18 hours with cytokines and the frequency of IFN-γ⁺ CD3⁻CD56^dim NK cells determined. Cytokines were IL-12, IL-15, or IL-12 and IL-15 (range: 0.001–100ng/ml). C and D. PBMC were cultured with Raji cells for 18 hours and then with K562 cells for 6 hours (E:T = 10:1). CD3⁻CD56^dim NK cells were assayed for CD107a (C) or IFN-γ (D). PBMC from all 60 donors were included in these analyses. M/M donors (dark blue boxes) M/T donors (light blue) and T/T donors (yellow). Statistical significance (p values) derives from two-tailed unpaired Mann Whitney tests (mean ± SE).

Figure 7. Substantial influences of -21 HLA-B dimorphism on NK cell education

A and B. In the ADCC assay, PBMC from 30 donors (10 M/M, 10 M/T and 10 T/T donors) were cultured for 6 hours with Raji cells (E:T = 10:1) coated with murine IgG (IgG) (squares) or with anti-CD20 Ab (circles) (2.5µg/ml). CD3⁻CD56^dim NK cells were assayed for IFN-γ (A) or CD107a (B). C. Using mass cytometry analysis and Boolean gating, the populations of IFN-γ⁺CD3⁻CD56^dim NK cells and CD107a⁺CD3⁻CD56^dim NK cells were each divided into eight subpopulations. Listed in the table, under the percentage of IFN-γ⁺ and CD107a⁺ NK cells, is the relative size of each subpopulation ± SE. Differences in the frequencies of these subpopulations in M/M and T/T donors were assessed with Mann-Whitney tests. Statistically significant differences are denoted with bold text. The gating strategy used to define the seven NK cell subsets is shown in fig. S9.

Figure 8. Substantial influence of -21 HLA-B dimorphism on the NK cell repertoire of expressed inhibitory HLA class I receptors

A. The frequencies of CD56^dim NK cells expressing C2-specifc KIR2DL1 in M/M, M/T and T/T donors. The color of a data point indicates the presence (green) or absence (black) of the C2 epitope in a donor. B. The frequencies of CD56^dim NK cells expressing C1-specifc KIR2DL2DL3 in M/M, M/T and T/T donors. The color of a data point indicates the presence (green) or absence (black) of the C1 epitope in a donor. C. The frequency in M/M (red bars) and T/T (yellow bars) of NK cells expressing the 64 phenotypic combinations of six inhibitory receptors that recognize HLA class I: LILRB1, KIR3DL2, KIR3DL1, KIR2DL2/3, KIR2DL1 and CD94:NKG2A. The asterisks indicate the 39 phenotypes that are present in all 60 donors. Data represent means (± SE) from 20 T/T and 20 M/M donors. p values are derived from unpaired Mann-Whitney tests to test for differences between M/M and T/T donors.