The MICA-129 dimorphism affects NKG2D signaling and outcome of hematopoietic stem cell transplantation

Abstract

The MHC class I chain-related molecule A (MICA) is a highly polymorphic ligand for the activating natural killer (NK)-cell receptor NKG2D. A single nucleotide polymorphism causes a valine to methionine exchange at position 129. Presence of a MICA-129Met allele in patients (n = 452) undergoing hematopoietic stem cell transplantation (HSCT) increased the chance of overall survival (hazard ratio [HR] = 0.77, P = 0.0445) and reduced the risk to die due to acute graft-versus-host disease (aGVHD) (odds ratio [OR] = 0.57, P = 0.0400) although homozygous carriers had an increased risk to experience this complication (OR = 1.92, P = 0.0371). Overall survival of MICA-129Val/Val genotype carriers was improved when treated with anti-thymocyte globulin (HR = 0.54, P = 0.0166). Functionally, the MICA-129Met isoform was characterized by stronger NKG2D signaling, triggering more NK-cell cytotoxicity and interferon-γ release, and faster co-stimulation of CD8(+) T cells. The MICA-129Met variant also induced a faster and stronger down-regulation of NKG2D on NK and CD8(+) T cells than the MICA-129Val isoform. The reduced cell surface expression of NKG2D in response to engagement by MICA-129Met variants appeared to reduce the severity of aGVHD.

Keywords: NK cells; NK‐cell receptors; cytotoxic T cells; graft‐versus‐host disease; single nucleotide polymorphism.

Figure 1. Cumulative survival according to MICA-129 genotype

1. Kaplan–Meier survival curves stratified by the patient MICA-129 genotype for all patients (n = 446). The survival curve is displayed for the first 66 months; 7.6% of the patients were followed longer. Effects on overall survival were determined by Cox regression with covariate adjustment as indicated in Table2. The HR indicates the risk per MICA-129Met allele carried by the patients (additive risk model). The numbers of patients carrying the three genotypes and the number of events (in brackets) are indicated.

2. Kaplan–Meier survival curves for patients receiving a graft matched for the MICA-129 genotype (n = 404).

3. Kaplan–Meier survival curves for patients who experienced aGVDH (any grade, n = 244).

4. Kaplan–Meier survival curves for patients not experiencing aGVHD (n = 189). The HR indicates the risk of patients carrying two MICA-129Met alleles (recessive risk model).

5. Kaplan–Meier survival curves for patients who did not receive a T-cell-depleting treatment with ATG (n = 193).

6. Kaplan–Meier survival curves for patients treated with ATG (n = 250).

7. Kaplan–Meier survival curves for patients with the MICA-129Val/Val genotype (n = 229) stratified by treatment with ATG.

8. Kaplan–Meier survival curves for patients carrying one or two MICA-129Met alleles (n = 214) stratified by treatment with ATG.

Figure 2. NKG2D binding to the MICA-129Met and MICA-129Val isoform and triggering of phosphorylation of SRC family kinases

1. The linear regression of MICA expression intensity and binding of a recombinant NKG2D-Fc fusion protein both determined as MFI by flow cytometry is displayed for L-MICA-129Met (n = 79, left panel) and L-MICA-129Val clones (n = 81, right panel). The coefficients of determination (R²), the regression coefficients (reg. coeff.), and the P-values for Pearson correlation are indicated.

2. Purified IL-2-stimulated (100 U/ml for 4 days) NK cells (10⁶) were stimulated with immobilized MICA-129Met-mIgG_2a-Fc or MICA-129Val-mIgG_2a-Fc or OVA-mIgG_2a-Fc fusion proteins (10 μg/ml) for 3, 10, or 30 min. The protein lysates of these cells were separated by SDS–PAGE, and the blot was probed subsequently with an anti-phospho-Tyr mAb, an anti-phospho-SRC family (Tyr419) kinases Ab, and an anti-β-actin mAb as a loading control. The arrow points toward phosphorylated SRC family kinases.

3. Blots obtained from three independent experiments were analyzed by densitometry, and the means plus SD of the ratio between phospho-SRC family kinase and β-actin signals is displayed. The difference between NK cells stimulated for 10 min by MICA-129Met-Fc or MICA-129Val-Fc proteins was assessed by t-test.

4. Purified IL-2-stimulated NK cells (100 U/ml for 4 days, 10⁶) were incubated with the SRC kinase inhibitor PP2 (25 μM), the vehicle DMSO, or medium only (Ø) for 30 min before being added to immobilized MICA-129Met-Fc, MICA-129Val-Fc, or OVA-Fc fusion proteins (10 μg/ml) for 10 min. The protein lysates of these cells were separated by SDS–PAGE, and the blot was probed subsequently with an anti-phospho-Tyr mAb and an anti-β-actin mAb as a loading control. The blot is representative for two independent experiments.

5. In parallel, degranulation of the NK cells was measured by anti-CD107a staining in flow cytometry. The difference between DMSO- and PP2-treated cells with respect to CD107a⁺ cells and the MFI of CD107a is indicated in the histograms. The results are representative for two independent experiments.

Figure EV1. NKG2D expression on freshly isolated and IL-2-stimulated NK-cell populations

1. Three populations (CD56^dimCD16⁺, CD56^brightCD16⁺, CD56^brightCD16⁻) were gated on freshly isolated NK cells (day 0) and NK cells stimulated for 4 days with IL-2 (100 U/ml) as exemplified in the upper panel. The NKG2 D expression on these NK-cell populations is displayed in the lower panel, and the MFI for NKG2D is indicated.

2. Summaries (means + SD) of the proportion of the three NK-cell populations and their NKG2D expression intensity (MFI and percentage NKG2D⁺ cells) at day 0 (n = 21) and 4 days after IL-2 stimulation (n = 16) are shown. The data were analyzed by t-test, and the P-values of significant differences are indicated.

Figure 3. Cytotoxic activity of NK cells in response to the MICA-129Met and MICA-129Val isoforms

1. The degranulation (CD107a expression) of purified IL-2-stimulated (100 U/ml for 4 days) NK cells in response to immobilized MICA-129Met-Fc and MICA-129Val-Fc, and as negative control, OVA-Fc fusion proteins was determined by flow cytometry after 1 h (n = 3). Displayed are means and SD of the MFI (left panel) and the percentage of CD107a⁺ cells (right panel). Differences between MICA-129Met-Fc- and MICA-129Val-Fc-induced NK-cell degranulation were analyzed by two-way ANCOVA adjusted for MICA protein concentration, and the respective P-values are indicated.

2. The degranulation of IL-2-stimulated LAK cells (100 U/ml for 4 days) exposed to L-MICA-129Met (n = 27) or L-MICA-129Val clones (n = 27) for 1 h was determined by flow cytometry. CD107a cell surface expression was analyzed after gating on CD56⁺ NK cells. In parallel, the MICA expression on target cells was determined. Displayed are the linear regressions of the MFI of CD107a on NK cells (upper panels) or the proportion of CD107a⁺ NK cells (lower panels) and the MICA expression intensity on target cells (MFI) for the L-MICA-129Met (left panels) and L-MICA-129Val clones (right panels). The coefficients of determination (R²), the regression coefficients (reg. coeff.), and the P-values for Pearson correlation are indicated.

3. A representative of 21 experiments is shown demonstrating the specific cytotoxic activity of LAK cells against an L-MICA-129Met and an L-MICA-129Val clone. L-con cells served as a negative and K562 cells as a positive control. The means of specific lysis of triplicates plus SD at different E:T ratios (200:1 to 3:1) were measured in an ⁵¹chromium-release assay. The MICA expression intensity and the binding of a recombinant NKG2D-Fc fusion protein to the target cells were determined in parallel by flow cytometry, and the MFIs are indicated.

Figure EV2. Degranulation of CD56^dimCD16⁺ and CD56^brightCD16⁺ NK cells in response to the MICA-129Met and the MICA-129Val isoforms

1. Purified IL-2-stimulated NK cells (100 U/ml for 4 days) were cultured for 2 h on plates coated with MICA-129Met-Fc, MICA-129Val-Fc, or OVA-Fc proteins (10 μg/ml) before flow cytometry. The NK-cell populations were gated as illustrated in the upper panel (CD56^dimCD16⁺, CD56^brightCD16⁺, CD56^brightCD16⁻), and CD107a expression was determined as displayed in the lower panel (red: CD107a, black: isotype control). The specific MFI (MFI CD107a minus MFI isotype control) and the percentage of CD107a⁺ cells are indicated.

2. A summary (means + SD) of 5 experiments is shown. The data were analyzed by t-test, and the P-values of significant differences are indicated.

Figure EV3. Expression of IFNγ in CD56^brightCD16⁻ and CD56^brightCD16⁺ NK cells in response to the MICA-129Met and the MICA-129Val isoforms

1. Purified IL-2-stimulated NK cells (100 U/ml for 4 days) were cultured for 4 h on plates coated with MICA-129Met-Fc, MICA-129Val-Fc, or OVA-Fc proteins (0, 0.1, 1, 5, 10, 15 μg/ml) before flow cytometry. The NK-cell populations were gated as exemplified in the upper panel (CD56^dimCD16⁺, CD56^brightCD16⁺, CD56^brightCD16⁻). The intracellular IFNγ expression in these populations is displayed in the lower panel (red: IFNγ, black: isotype control). The specific MFI (MFI IFNγ minus MFI isotype control) and the percentage of IFNγ⁺ cells are indicated.

2. A summary (means + SEM) of 9 experiments is shown. The data were analyzed by two-way ANCOVA adjusted for the protein concentration (5, 10, 15 μg/ml), and the P-values of significant differences between the MICA-129Met-Fc and MICA-129Val-Fc proteins are indicated.

Figure 4. IFNγ production of NK cells in response to the MICA-129Met and MICA-129Val isoforms

1. The IFNγ expression of purified IL-2-stimulated (100 U/ml for 4 days) NK cells exposed for 4 h to L-MICA-129Met (n = 23) or L-MICA-129Val clones (n = 23) was determined by flow cytometry. The IFNγ expression was analyzed after gating on CD56^brightCD16⁻ NK cells. In parallel, the MICA expression on target cells was determined. Displayed are the linear regressions of the MFI of IFNγ in CD56^brightCD16⁻ NK cells (upper panels) or the proportion of IFNγ⁺ CD56^brightCD16⁻ NK cells (lower panels) and the MICA expression intensity on target cells (MFI) for the L-MICA-129Met (left panels) and L-MICA-129Val clones (right panels). The coefficients of determination (R²), the regression coefficients (reg. coeff.), and the P-values for Pearson correlation are indicated.

2. The IFNγ release of purified IL-2-stimulated NK cells (100 U/ml for 4 days) co-cultured with L-MICA-129Met (n = 34) or L-MICA-129Val clones (n = 32) for 24 h was measured in the supernatant by ELISA. In parallel, the MICA expression intensity on target cells was determined by flow cytometry. The linear regressions of IFNγ release (pg/ml) by NK cells and MICA expression on targets (MFI) are displayed for the L-MICA-129Met clones (left panel) and the L-MICA-129Val clones (right panel). The coefficients of determination (R²), the regression coefficients (reg. coeff.), and the P-values for Pearson correlation are indicated.

Figure 5. Down-regulation of NKG2D on NK cells in response to the MICA-129Met and MICA-129Val isoforms

NKG2D expression on purified IL-2-stimulated NK cells (100 U/ml for 4 days) exposed to L-MICA-129Met (n = 25) or L-MICA-129Val clones (n = 25) for 0, 4, and 24 h was analyzed by flow cytometry. NK cells (2.5 × 10⁵) were co-cultured with 5 × 10⁴ target cells and analyzed for NKG2D expression after gating of CD3⁻CD56⁺ cells. The means and SD of the percentage of NKG2D⁺ NK cells (left panel) and of the MFI of NKG2D (right panel) are shown. Differences between the groups were analyzed by repeated measures ANOVA, and P-values for pairwise comparisons are indicated.

Figure EV4. Down-regulation of NGK2D on CD56^dimCD16⁺, CD56^brightCD16⁺, and CD56^brightCD16⁻ NK cells in response to the MICA-129Met and the MICA-129Val isoforms

1. Purified IL-2-stiumlated NK cells (100 U/ml for 4 days) were co-cultured with L-con, L-MICA-129Met, or L-MICA-129Val clones for 4 h. Three NK-cell populations (CD56^dimCD16⁺, CD56^brightCD16⁺, CD56^brightCD16⁻) were gated as illustrated in the upper panel. The NKG2D expression on these NK-cell populations is displayed in the lower panel, and the MFI of NKG2D and the percentages of NKG2D⁺ cells are indicated.

2. A summary (means + SD) of NKG2D expression on the three NK-cell populations 4 and 24 h after co-culture with L-con (n = 3), L-MICA-129Met (n = 17), and L-MICA-129Val clones (n = 18) is displayed. The NKG2D expression (MFI) at the beginning (0 h) was set to 100%. The average reduction of NKG2D on NK cells in response to L-MICA-129Met compared to L-MICA-129Val clones at 4 and 24 h (%-points) is indicated in the panels. The differences were analyzed by repeated measures ANOVA, and P-values are indicated.

Figure 6. Co-stimulation of CD8⁺ T cells by the MICA-129Met and MICA-129Val isoforms

1. MACS-separated CD8⁺ T cells were cultured in triplicate on an immobilized anti-CD3 mAb (0.005 μg/ml [upper panel] or 0.01 μg/ml [lower panel]) in combination with recombinant MICA-129Met-Fc, MICA-129Val-Fc, and OVA-Fc proteins at various concentrations (1.0, 0.5, 0.1, 0.0 μg/ml). After 72 h, 25% of the supernatant was harvested and IL-2 concentrations were measured by ELISA. The harvested medium was replaced by the same volume containing 1 μCi ³H-labeled thymidine. After 12 h, the plates were completely harvested and the DNA-bound radioactivity was determined. The means and SD of the stimulation index (SI) are displayed (n = 4). Significant differences between MICA-129Met/Val-Fc and OVA-Fc proteins were found when the antigen-specific signal (anti-CD3) was limited (*P < 0.05, t-test; upper left panel: 1.0 μg/ml: MICA-129Met-Fc versus OVA-Fc P = 0.0372 and MICA-129Val-Fc versus OVA-Fc P = 0.0366; upper right panel: 1.0 μg/ml: MICA-129Met-Fc versus OVA-Fc P = 0.0499 and MICA-129Val-Fc versus OVA-Fc P = 0.0192; 0.5 μg/ml: MICA-129Met-Fc versus OVA-Fc P = 0.0164 and MICA-129Val-Fc versus OVA-Fc P = 0.0357; lower left panel: 0.5 μg/ml: MICA-129Met-Fc versus OVA-Fc P = 0.0287 and MICA-129Val-Fc versus OVA-Fc P = 0.0232; lower right panel: 1.0 μg/ml: MICA-129Met-Fc versus OVA-Fc P = 0.0171 and MICA-129Val-Fc versus OVA-Fc P = 0.0484).

2. Purified CFSE-stained CD8⁺ T cells were stimulated by immobilized anti-CD3 (0.005 μg/ml) in combination with recombinant MICA-129Met-Fc, MICA-129Val-Fc, OVA-Fc proteins, or co-stimulatory mAb (anti-CD28, anti-NKG2D) or an isotype control (mIgG₁). The proliferation of CD3⁺CD8⁺ T cells was assessed at 60 h by flow cytometry. Results of a representative out of 6 experiments are displayed. Untreated CFSE-stained CD8⁺ T cells are included for comparison. The percentage of proliferating cells and MFI for CFSE are indicated.

3. The MFI of CFSE in unstimulated CD8⁺ T cells (control) was set to 100% in individual experiments (n = 6), and the relative decrease due to proliferation was calculated. Means + SD are shown. Significant differences (*P = 0.0277, Wilcoxon test) between MICA-129Met-Fc versus MICA-129Val-Fc and OVA-Fc proteins were found at slightly higher anti-CD3 concentrations (0.1 and 0.05 μg/ml) than at later time points (see A).

4. Anti-CD28 and anti-NKG2D mAb were used in parallel as a positive control, mean + SD are shown, and significant differences (*P = 0.0277, Wilcoxon test) to the isotype control (mIgG₁) are indicated (n = 6).

Figure EV5. Down-regulation of NGK2D on CD8⁺ T cells in response to the MICA-129Met and MICA-129Val isoforms

1. Purified CD8⁺ T cells were analyzed by flow cytometry for NKG2D expression as illustrated here. The MFI for NKG2D and the percentage of NKG2D⁺CD8⁺ T cells are indicated.

2. The NK cells were subsequently co-cultured with an L-con, L-MICA-129Met, or L-MICA-129Val clone (the MFI values for MICA on these clones are indicated above the histograms in brackets). NKG2D expression was determined as illustrated in (A) after 4 and 24 h. The MFI for NKG2D and the percentages of NKG2D⁺CD8⁺ T cells are indicated.

Figure 7. Down-regulation of NKG2D by the MICA-129Met and the MICA-129Val isoforms on CD8⁺ T cells and impairment of subsequent co-stimulation via NKG2D

1. NKG2D expression on purified CD8⁺ T cells exposed to L-MICA-129Met (n = 19) or L-MICA-129Val clones (n = 19) for 0, 4, and 24 h was analyzed by flow cytometry. CD8⁺ T cells (2.5 × 10⁵) were co-cultured with 5 × 10⁴ target cells and analyzed for NKG2D expression after gating on CD3⁺CD8⁺ T cells. The means and SD of the MFI of NKG2D (left panel) and of the percentage of NKG2D⁺CD8⁺ T cells (right panel) are displayed. Differences between the groups were analyzed by repeated measures ANOVA, and the P-values are indicated.

2. Purified CD8⁺ T cells were cultured on plate-bound anti-NKG2D (1 μg/ml) or isotype control (mIgG₁) for 24 h before the NKG2D expression was measured by flow cytometry. Means and SD of MFI (upper panel) and percentage of NKG2D⁺ cells (lower panel) are shown (n = 6). Differences between the groups were analyzed by t-tests, and the P-values are indicated.

3. These CD8⁺ T cells were subsequently CFSE-stained and cultured on plates coated with anti-CD3 (0.005 μg/ml) in combination with anti-CD28 (0.5 μg/ml) as a positive control or anti-NKG2D (0.5 μg/ml). Proliferation was measured after 60 h by flow cytometry. Untreated CFSE-stained CD8⁺ T cells are included for comparison. The percentage of proliferating cells and the MFI for CFSE are indicated.

4. The MFI of CFSE in unstimulated CD8⁺ T cells (control) was set to 100% in individual experiments (n = 6), and the relative decrease due to proliferation was calculated. Means + SD are shown. Significant differences (Wilcoxon test) between CD8⁺ T cells pre-exposed to anti-NKG2D and isotype control (mIgG₁) were found in these experiments at anti-CD3 concentrations of 0.01 and 0.005 μg/ml.

Figure 8. Summary of functional effects of MICA-129 variants depending on expression intensity

1. For target cells expressing the MICA-129Val variant, the degree of NK-cell cytotoxicity and IFNγ production increased steadily with the MICA expression intensity. Augmented expression of the high-avidity MICA-129Met isoform, in contrast, had no or even a negative effect on these NK-cell functions due to a rapid down-regulation of NKG2D on NK cells.

2. Antigen-dependent co-stimulation of CD8⁺ T cells with the MICA-129Met variant allowed for an earlier antigen-dependent activation than co-stimulation with the MICA-129Val variant. However, the down-regulation of NKG2D in response to MICA-129Met ligands impaired any subsequent NKG2D-dependent co-stimulation and T-cell activation. The down-regulation of NKG2D on CD8⁺ T cells was augmented with MICA-129Met expression intensity.