Adaptive Natural Killer Cells Integrate Interleukin-18 during Target-Cell Encounter

Abstract

Human cytomegalovirus (HCMV) infection induces adaptations in the natural killer (NK)-cell compartment. Expanded subsets of adaptive NK cells display potent effector functions against cellular targets, despite their apparent unresponsiveness to stimulation with classical dendritic cell-derived cytokines interleukin (IL)-12 and IL-18. However, it remains unclear whether adaptive NK cells have completely lost their ability to sense inflammation via IL-12 and IL-18 or whether these pro-inflammatory signals can be functionally integrated into defined contexts. Here, we demonstrate that adaptive NKG2C⁺ NK cells can be costimulated by the presence of pro-inflammatory cytokines during target cell-induced activation. Cytokine costimulation of adaptive NK cells resulted in elevated interferon (IFN)-gamma and tumor necrosis factor (TNF) production, which promoted protein expression of HLA class I and adhesion molecules as well as transcription of genes involved in antigen processing and antiviral states in endothelial bystander cells in vitro. We further show that IL-18 drove costimulation in functional assays and was sufficient for elevated cytokine production in the absence of IL-12. Hence, adaptive NKG2C⁺ NK cells-although poorly responsive to IL-12 and IL-18 as an isolated stimulus-integrate IL-18 as a costimulatory signal during target-cell encounter.

Keywords: NKG2C; adaptive natural killer cells; costimulation; human natural killer cells; interleukin-18.

Figure 1

Effector responses of adaptive natural killer (NK) cells against target cells are amplified by cytokine costimulation. (A) Representative staining of interferon (IFN)-γ gated on conventional NKG2C⁻ or adaptive NKG2C⁺ NK cells after 24 h culture in the absence or presence of interleukin (IL)-12 + 18. (B) Summary of frequencies of IFN-γ⁺ cells. Symbols indicate individual donors, and red lines indicate median (n = 9). (C) Representative staining of IFN-γ gated on adaptive NKG2C⁺ NK cells after 6 h culture either alone or with K562/HLA-E target cells in the absence or presence of IL-12 + 18. (D) Summary of frequencies of IFN-γ⁺ cells and (E) summary of geometric mean fluorescence intensity (geoMFI) of IFN-γ in IFN-γ⁺ cells, n = 17 donors. (F) Summary of IFN-γ protein concentrations in the supernatant of cocultures, n = 7 donors. (G) Representative staining of tumor necrosis factor (TNF) gated on adaptive NKG2C⁺ NK cells after 6 h culture either alone or with K562/HLA-E target cells in the absence or presence of IL-12 + 18. (H) Summary of frequencies of TNF⁺ cells and (I) summary of geoMFI of TNF in TNF⁺ cells, n = 17 donors. (J) Summary of TNF protein concentrations in the supernatant of cocultures, n = 7 donors. Connected symbols indicate individual donors, and red lines indicate median. All statistical analyses performed with one-tailed Wilcoxon matched-pairs test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Figure 2

Cytokine costimulated adaptive natural killer (NK) cells proficiently alert bystander cells in vitro. (A) Schematic illustration of experimental design. (B) Representative geoMFI of HLA class I on human umbilical vein endothelial cells (HUVEC) after 40 h treatment with indicated conditioned medium in the presence or absence of anti-interferon (IFN)-γ and anti-tumor necrosis factor (TNF) blocking antibodies (left) and summary of n = 8 treatments with conditioned medium from independent NK-cell stimulations (right). (C) Representative frequency of E-selectin⁺ HUVEC after 6 h treatment with indicated conditioned medium in the presence or absence of anti-TNF blocking antibodies (left) and summary of n = 8 treatments with conditioned medium from independent NK-cell stimulations (right). (D) Representative PSMB9 transcript abundance relative to GAPDH in HUVEC after 24 h treatment with indicated conditioned medium (left) and summary of n = 8 NK-cell stimulations (right). (E) Representative protein kinase R (PRK) PKR transcript abundance relative to GAPDH in HUVEC after 24 h treatment with indicated conditioned medium (left) and summary of n = 8 treatments with conditioned medium from independent NK-cell stimulations (right). Connected symbols indicate individual donors used to generate conditioned medium. All statistical analyses performed with one-tailed Wilcoxon matched-pairs test. NS, not significant; *p < 0.05 and **p < 0.01.

Figure 3

Interleukin (IL)-18 drives costimulation of adaptive natural killer (NK) cells. (A) IL12RB2 and (B) IL18RAP transcript abundance relative to GAPDH in ex vivo FACS-purified immature conventional NK cells (CD56^dim CD57⁻ NKG2C⁻), mature conventional NK cells (CD56^dim CD57⁺ NKG2C⁻), adaptive NKG2C⁺ NK cells (CD56^dim CD57⁺ NKG2C⁺), naïve T cells (CD3⁺ CD8⁺ CD45RA⁺ CD45RO⁻ CCR7⁺), and effector-memory T cells (CD3⁺ CD8⁺ CD45RA⁻ CD45RO⁺ CCR7⁻), n = 6 individual donors. Box plots display minimum to maximum and median. Statistical analysis performed with Friedman and Dunn’s multiple comparison test. (C) Representative staining of NF-κB(pS529) in FACS-purified adaptive NKG2C⁺ NK cells after treatment with medium, phorbol-12-myristat-13-acetat (PMA), or interleukin (IL)-18 for 15 min. (D) Summary of frequencies of NF-κB(pS529)⁺ cells either treated with medium or IL-18 over time, n = 5 donors. Symbols indicate mean and error bars SEM. Statistical analysis performed with repeated-measures two-way ANOVA with Bonferroni correction. (E,F) NK cells were cultured with K562/HLA-E in the presence of IL-12 + 18, IL-12, or IL-18. (E) Frequencies of ΔIFN-γ⁺ [calculated by subtraction of the frequency of interferon (IFN)-γ⁺ cells in the presence of K562/HLA-E alone] and (F) ΔTNF⁺ adaptive NK cells, n = 12 donors. Symbols indicate mean and error bars SEM. Statistical analysis performed with Friedman and Dunn’s multiple comparison test. (G,H) NK cells were cultured with K562/HLA-E and varying concentrations of IL-18. (G) Frequencies of ΔIFN-γ⁺ and (F) ΔTNF⁺ adaptive NK cells, n = 6 donors. Symbols indicate mean and error bars SEM. Statistical analysis performed with Friedman and Dunn’s multiple comparison test. NS, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.