Supplementary Figure 1: Sequence variations in HCMV UL40-encoded peptides control the activation of adaptive NKG2C⁺ NK cells but do not differentially affect inhibition of NKG2C^– NKG2A⁺ NK cells.

(a-b) PBMC of healthy HCMV^– (n=20) and HCMV⁺ (n=40) donors were screened by flow cytometry. (a) Frequency of NKG2C⁺ cells within the CD56^dim population and (b) frequency of CD2⁺ Siglec-7^– NKG2A^– FcεR1γ^– cells within the CD56^dim NKG2C⁺ population. Symbols indicate individual donors and lines median. CV, coefficient of variation. (c) Gating strategy for functional assays using HCMV⁺ donors with adaptive NKG2C⁺ NK cells. After culture of purified viable CD3^– CD56⁺ NK cells with peptide-pulsed target cells, adaptive NKG2C⁺ NK cells were gated as viable single CD56^dim NKG2A^– CD57⁺ KIR⁺ NKG2C⁺ cells. Depending on the phenotype of the individual donor, KIR were gated as KIR2DL1⁺, KIR2DL3⁺, or KIR3DL1⁺. (d) Purified NK cells from HCMV⁺ donors were used as effector cells in cytotoxicity assays against labelled peptide-pulsed RMA-S–HLA-E cells and cytotoxicity was calculated as described in the Methods section. Symbols indicate individual data points, lines indicate means, and error bars indicate SEM (n=6 individual donors in 3 independent experiments). Two-way repeated-measures ANOVA with Bonferroni correction between VMAPRTLIL and VMAPRTLFL. (e) RMA-S–HLA-E cells were pulsed with 300 μM of the indicated peptides and geometric mean fluorescence intensity (geoMFI) of HLA-E surface expression was assed. Symbols indicate independent experiments (n=6) and horizontal lines median. Friedman test with Dunn’s post test. (f) Binding affinities were predicted using the NetMHC4.0 algorithm. The HCMV pp65-derived HLA-A2-restricted NLVPMVATV peptide served as a non-HLA-E-binding control. (g) RMA-S–HLA-E cells were pulsed with 300 μM VMAPRTLIL or VMAPRTLFL peptide followed by removal of peptide and chase for 6 h. Decay in HLA-E surface expression was calculated assuming first order kinetics. Symbols indicate individual data points, error bars indicate SEM (n=3 independent experiments), and lines indicate linear regression curves. Slopes of regression curves were compared using ANCOVA. (h) RMA-S–HLA-E cells were pulsed with increasing concentrations of the indicated peptides and geoMFI of HLA-E surface expression upon pulsing was assessed. Symbols indicate individual data points, lines indicate means, and error bars indicate SEM (n=6 independent experiments). Two-way repeated-measures ANOVA with Bonferroni correction between VMAPRTLIL and VMAPRTLFL. (i) Degranulation response of viable CD56^dim NKG2C^– (triangles) or viable CD56^dim NKG2A^– CD57⁺ KIR⁺ NKG2C⁺ NK cells (circles) upon culture without or with VMAPRTLFL-pulsed RMA-S–HLA-E cells. Connected symbols represent individual donors (n=12 in 6 experiments). Two-tailed Wilcoxon test. (j) Sorted viable CD56^dim NKG2A^– NKG2C⁺ NK cells from HCMV⁺ donors were treated with IgG1 isotype control or anti-CD94 blocking antibody prior to culture without or with VMAPRTLFL-pulsed RMA-S–HLA-E cells. Summary of degranulation of viable CD56^dim NKG2A^– CD57⁺ NKG2C⁺ NK cells. Connected symbols represent individual donors (n=6 in 3 independent experiments). Two-tailed Wilcoxon test. (k) Purified NK cells from HCMV⁺ donors were cultured with K562–HLA-E cells pulsed with indicated peptides at indicated concentrations. Summary of CCL3 expression and degranulation gated on viable CD56^dim NKG2A^– CD57⁺ KIR⁺ NKG2C⁺ NK cells (circles) or CD56^dim NKG2C^– NKG2A⁺ cells (triangles). Symbols indicate individual data points, lines indicate means, and error bars indicate SEM (n=6 individual donors in 3 independent experiments). Two-way repeated-measures ANOVA with Bonferroni correction between VMAPRTLIL and VMAPRTLFL. NS not significant, *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001.