Splenic TNF-α signaling potentiates the innate-to-adaptive transition of antiviral NK cells

Abstract

Natural killer (NK) cells possess both innate and adaptive features. Here, we investigated NK cell activation across tissues during cytomegalovirus infection, which generates antigen-specific clonal expansion and long-lived memory responses. Longitudinal tracking and single-cell RNA sequencing of NK cells following infection revealed enhanced activation in the spleen, as well as early formation of a CD69^lo precursor population that preferentially gave rise to adaptive NK cells. Splenic NK cells demonstrated heightened tumor necrosis factor alpha (TNF-α) signaling and increased expression of the receptor TNFR2, which coincided with elevated TNF-α production by splenic myeloid cells. TNFR2-deficient NK cells exhibited impaired interferon gamma (IFN-γ) production and expansion. TNFR2 signaling engaged two distinct nuclear factor κB (NF-κB) signaling arms-innate effector NK cell responses required canonical NF-κB signaling, whereas non-canonical NF-κB signaling enforced differentiation of CD69^lo adaptive NK cell precursors. Thus, NK cell priming in the spleen during viral infection promotes an innate-to-adaptive transition, providing insight into avenues for generating adaptive NK cell immunity across diverse settings.

Keywords: MCMV; NF-κB signaling; NIK; NK cells; TNF-α; TNFR2.

Figure 1.. Spleen facilitates adaptive NK cell expansion and formation of a CD69^lo adaptive NK cell precursor population during MCMV infection

A) Frequency of splenic Ly49H⁺ NK cells that were adoptively transferred into Klra8^−/− mice and recovered from indicated organs 5 days p.i. B) UMAP display of Ly49H⁺ NK cells isolated 4 days p.i. for single cell RNA-sequencing analysis (from Supplementary Fig. 1I) that underwent reclustering analysis (left) and with cell proliferation score plotted (right). C) Heatmap of selected genes from selected pathways identified from Gene Set Enrichment Analysis enriched in Proliferation^hi clusters (clusters 1, 4, 5) or enriched in Proliferation^lo clusters (cluster 0, 2). D) Frequency of wildtype Ly49H⁺ NK cells that express CD69 following MCMV infection (left) with representative flow cytometry plot of CD69 expression 4 days p.i. (right). E) Frequency or surface levels of specified receptors amongst wildtype spleen CD69^hi and CD69^lo NK cells 4 days p.i. F) Frequency of wildtype spleen CD69^hi and CD69^lo Ly49H⁺ NK cells that express KLRG1 (left) or CD27 and CD11b (right) 4 days p.i. G) Frequency of spleen CD69^hi and CD69^lo Ly49H⁺ NK cells that are BIM^hi 4 days p.i. (left). H) Frequency of spleen CD69^hi and CD69^lo Ly49H⁺ NK cells that have incorporated EdU 4 days p.i. I) CD69^hi and CD69^lo Ly49H⁺ NK cells were sorted from congenically-distinct wildtype spleens 4 days p.i. and adoptively co-transferred into infection-matched Klra8^−/− mice that had received wildtype bystander NK cells prior to infection. Frequency of Ly49H⁺ NK cells that were originally CD69^hi or CD69^lo and recovered from the spleens of recipient mice 7 days p.i. was quantified. Data are representative of at least two (E-I) or three (A,D) independent experiments. Each symbol represents the group mean of 3–5 mice (A,D,I) or an individual mouse (E-H). ****P<0.0001, ***P<0.001 **P<0.01 with two-way ANOVA (F) or paired t-test (E-I). Error bars, mean ± s.e.m. (A, D-H). See also Figure S1.

Figure 2.. Spleen supports adaptive NK cell priming early during MCMV infection

A) Frequency of Ly49H⁺ NK cells that express CD69⁺ in the spleen or liver from wildtype mice (left) with representative histogram 4 days p.i. (right). B) Frequency of Ly49H⁺ NK cells that have incorporated EdU in the spleen or liver from wildtype mice that were uninfected or 4 days p.i C) Equivalent numbers of Ly49H⁺ NK cells were sorted from wildtype mouse spleen or liver 1 day p.i. and co-transferred into infection-matched Klra8^−/− mice before recovery 6 days later. Frequency of Ly49H⁺ NK cells that originated from a given priming site was quantified. D) Spleen and liver Ly49H⁺ NK cells from wildtype mice were sorted 1 day p.i. and processed for RNA-sequencing analysis. Gene-set enrichment analysis of genes differentially expressed (DE) in splenic Ly49H⁺ NK cells compared to DE genes in liver Ly49H⁺ NK cells was performed with selected plots of spleen-enriched pathways displayed. E) Intracellular MYC protein levels expressed by spleen or liver Ly49H⁺ NK cells (left) from wildtype mice with representative histogram 2 days p.i. (right). Data are representative of at least two (B, C, E) or three (A) independent experiments. Each symbol represents the group mean of 3–5 mice (A, E) or an individual mouse (B-C). ****P<0.0001, ***P<0.001 with two-way ANOVA test. Error bars, mean ± s.e.m. (A-C,E). See also Figure S2.

Figure 3.. Rapid TNF-α-TNFR2 signaling initiated during viral infection

A) Top 5 spearman coefficients between log2FC of differentially accessible (DA) peaks from ATAC-seq analysis and log2FC of differentially expressed (DE) genes from RNA-seq analysis. Ranked on coefficient value. B) Scatter plot of genes in “Tumor Necrosis Factor Pathway” classification. Genes that were DE between 0 and 2 days p.i. (d2/d0) during Ly49H⁺ NK cell MCMV response are displayed in red. C) RNA-seq analysis of Tnfrsf1b levels in spleen Ly49H⁺ NK cells at steady-state and 2 days p.i.. D) RNA-seq analysis of Tnfrsf1b levels in NK cells that were exposed to IFN-α or IL-12 and IL-18 in vitro for 3 hours. E) RNA-seq analysis of Tnfrsf1b expressed by wildtype or Il12rb2-deficient Ly49H⁺ NK cells isolated from mixed bone marrow chimeras at steady-state or 2 days p.i.. F) RNA-seq analysis of Tnfrsf1b expressed by wildtype or Stat4-deficient Ly49H⁺ NK cells isolated from mixed bone marrow chimeras at steady-state or 2 days p.i.. G) Kaplan-Meier survival curve of wildtype or TNFR2^−/− mice following MCMV infection. H) Protein levels of TNF-α in homogenized spleen or liver tissues in wildtype mice following MCMV infection. I) Absolute numbers of TNF-α⁺ cells detected by flow cytometry in the spleen or liver of wildtype mice 2 days p.i. Each symbol represents the group mean of 2–5 mice (D, H) or an individual mouse (C, E-F, I). ****P<0.0001, **P<0.01 with two-way ANOVA test (H-I) or Mantel-Cox log-rank test (G). *Padj<0.05 from RNA-seq analysis (C-F). Error bars, mean ± s.e.m. (C-F, H-I). See also Figure S3.

Figure 4.. TNFR2 induces innate effector functions of NK cells through canonical NF-κB and p38 signaling

A) Frequency of Ly49H⁺ spleen NK cells that are IFN-γ⁺ following anti-TNF-α blockade in wildtype mice 2 days p.i. B) Frequency of Ly49H⁺ wildtype or TNFR2-deficient spleen NK cells in mixed bone marrow chimera mice that are IFN-γ⁺ 2 days p.i. C) Frequency of spleen NK cells that are IFN-γ⁺ following exposure to IL-12 and/or TNF-α for 4–5 hours. D) Frequency of spleen NK cells that are IFN-γ⁺ following exposure to IL-12 and/or TNF-α for 4–5 hours and treatment with anti-TNFR1 and/or anti-TNFR2 blocking antibodies. E) Schematic of TNFR2 signaling. F) Frequency of spleen NK cells that are IFN-γ⁺ following exposure to IL-12 and/or TNF-α for 5 hours and treatment with indicated inhibitors at varying concentrations. G) Frequency of wildtype or NIK^−/− NK cells from the spleens of mixed bone chimera mice that are IFN-γ⁺ following exposure to IL-12 and/or TNF-α for 5 hours. Data are representative of at least two (D, F-G) or three (A-C) independent experiments. Each symbol represents individual mice (A-B, G) or the group mean of 3 mice (C-D, F). ***P<0.001 **P<0.01, *P<0.05 with one-way ANOVA between control and other conditions (C-D, F), two-way ANOVA (G), or unpaired (A) or paired t-test (B). Error bars, mean ± s.e.m. (A-D, F-G). See also Figure S4.

Figure 5.. TNFR2 enhances adaptive NK cell responses in the spleen

A) Quantification of MYC levels (left), CD71 frequency (middle), and CD25 frequency (right) amongst wildtype and TNFR2-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice 2 days p.i. B) Frequency of wildtype and TNFR2-deficient Ly49H⁺ NK cells that had been adoptively co-transferred to Klra8^−/− mice and recovered from their spleens 4–7 days p.i. (left) or indicated organs 5 days p.i. (right). C) Heatmap of DE genes from RNA-seq analysis between wildtype and TNFR2-deficient Ly49H⁺ NK cells from mixed bone marrow chimera mice following MCMV infection. Selected genes labeled. D) Display of selected pathways from analysis of DE genes between wildtype and TNFR2-deficient Ly49H⁺ NK cells from mixed bone marrow chimera mice 4 days p.i. E) Representative histogram (top) and quantified frequency (bottom) of CD69 (left), CD38 (middle), or TOX (right) in wildtype or TNFR2-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice 5 days p.i. Uninfected control in gray shading. F) Representative histogram (left) and quantified frequency (right) of BIM^hi wildtype or TNFR2-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice 5 days p.i. G) Frequency of wildtype or TNFR2-deficient spleen Ly49H⁺ NK cells from mixed bone chimera chimera mice in a given cell cycle stage 5 days p.i. as determined by EdU incorporation and FxCycle levels (Supplementary Fig. 5G). Data are representative of at least two independent experiments (A-B, E-G). Each symbol represents individual mice (A-B, E-G). ****P<0.0001, ***P<0.001, **P<0.01 with two-way ANOVA (A-B, G) or paired t-test (E-F) or. Error bars, mean ± s.e.m. (A-B, E-G). See also Figure S5.

Figure 6.. Noncanonical NF-kB signaling mediates adaptive NK cell expansion

A) Quantification of Nfkb2 and Relb levels following isolation and RNA-seq analysis of wildtype and TNFR2-deficient spleen NK cells from mixed bone chimera mice 4 days p.i. B) Frequency of wildtype and NIK-deficient Ly49H⁺ NK cells from mixed bone marrow chimera spleens that were adoptively co-transferred to Klra8^−/− mice and recovered from their spleens 4–7 days (left) or indicated organs 5 days (right) p.i. C) Representative histogram (top) and quantified frequency (bottom) of CD69 (left), CD38 (middle), or TOX (right) in wildtype or NIK-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice 5 days p.i. Uninfected control in gray shading. D) Representative histogram (left) and quantified frequency (right) of BIM^hi wildtype or NIK-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice 5 days p.i. E) Frequency of wildtype or NIK-deficient spleen Ly49H⁺ NK cells from mixed bone marrow chimera mice that are in a given cell cycle stage 5 days p.i. as determined by EdU incorporation and FxCycle levels (Supplementary Fig. 5G). Data are representative of at least two independent experiments (B-E). Each symbol represents individual mice (A-E). *Padj<0.05 from RNA-seq analysis (A); ****P<0.0001, **P<0.01 with paired t-test (C-D) or two-way ANOVA test (B,E). Error bars, mean ± s.e.m. (A-E). See also Figure S6.