Differential Induction of IFN-α and Modulation of CD112 and CD54 Expression Govern the Magnitude of NK Cell IFN-γ Response to Influenza A Viruses

Abstract

In human and murine studies, IFN-γ is a critical mediator immunity to influenza. IFN-γ production is critical for viral clearance and the development of adaptive immune responses, yet excessive production of IFN-γ and other cytokines as part of a cytokine storm is associated with poor outcomes of influenza infection in humans. As NK cells are the main population of lung innate immune cells capable of producing IFN-γ early in infection, we set out to identify the drivers of the human NK cell IFN-γ response to influenza A viruses. We found that influenza triggers NK cells to secrete IFN-γ in the absence of T cells and in a manner dependent upon signaling from both cytokines and receptor-ligand interactions. Further, we discovered that the pandemic A/California/07/2009 (H1N1) strain elicits a seven-fold greater IFN-γ response than other strains tested, including a seasonal A/Victoria/361/2011 (H3N2) strain. These differential responses were independent of memory NK cells. Instead, we discovered that the A/Victoria/361/2011 influenza strain suppresses the NK cell IFN-γ response by downregulating NK-activating ligands CD112 and CD54 and by repressing the type I IFN response in a viral replication-dependent manner. In contrast, the A/California/07/2009 strain fails to repress the type I IFN response or to downregulate CD54 and CD112 to the same extent, which leads to the enhanced NK cell IFN-γ response. Our results indicate that influenza implements a strain-specific mechanism governing NK cell production of IFN-γ and identifies a previously unrecognized influenza innate immune evasion strategy.

Figure 1.. NK cells suppress viral infection and secrete IFN-γ in response to pandemic 2009 H1N1 influenza A virus.

(A) Diagram of gating tree used to identify monocytes and NK cells by flow cytometry. FSC, forward scatter; AViD, Live/Dead fixable aqua dead cell stain; SSC-A, side scatter. Cal/09 infection at an MOI = 3 at 7 HPI is shown. (B) Representative flow plot of CD107a expression on NK cells after 6 or 23-hr co-culture with Cal/09-infected monocytes (MOI = 3). (C) Summary plot of NK cell CD107a⁺ frequency after 6 or 23-hr (n = 19) co-culture with Cal/09-infected monocytes. PMA/ionomycin treatment served as positive control. (D) The percent dead monocytes of mock (n = 7) or Cal/09 (n = 16) monocytes alone (M) or co-cultured with NK cells at an E:T of 1:1 (M+NK). (E) Representative flow plot of NK cell IFN-γ⁺ frequency after after 6 or 23-hr co-culture with Cal/09-infected monocytes. (F) Summary plot of NK cell IFN-γ⁺ frequency after 6 or 23-hr (n = 12) co-culture with Cal/09-infected monocytes or exposure to virions at 7 or 24 HPI assessed by intracellular cytokine staining. PMA/ionomycin treatment served as positive control. *P < 0.05, Wilcoxon signed-rank test. (G) Summary plot of NK cell IFN-γ⁺ frequency after 23-hr co-culture with Cal/09-infected monocytes incubated with anti-IL-2 or an isotype control antibody (4 µg/mL) and assessed by intracellular cytokine staining (n = 6).

Figure 2.

NK cells suppress viral infection and secrete IFN-γ in response to six influenza A strains. (A) The percent Flu-NP⁺ monocytes of indicated influenza strain 24 HPI (MOI = 3) and an E:T ratio of 0:1 without NK cells (M) or with NK cells at an E:T of 1:1 (M+NK) (n = 6). (B) The percent Flu-NP⁺ monocytes measured after Cal/09 or Vic/11 infection at an MOI of 3 and an E:T ratio of 0:1, 1:4, 1:1 or 4:1 for a total of 7 (n = 7) or 24 hr (n = 9). (C) RNA expression levels of an identical sequence in the influenza matrix gene segment of Cal/09 and Vic/11 normalized to the housekeeping gene GAPDH (n = 3). (D) NK cell IFN-γ⁺ frequency after culture with monocytes infected with the specified influenza A strains at 7 HPI (n = 7) or 24 HPI (n = 6). (E) IFNG transcript levels normalized to GAPDH in the monocyte alone or monocyte – NK cell co-culture (M+NK) after mock treatment, Cal/09 or Vic/11 exposure (MOI = 3). Values represent average of technical triplicates (n = 3). (F) Representative flow plot showing percent IFN-γ⁺ of cord blood NK cells post exposure to mock, Cal/09- or Vic/11-infected monocytes (n = 3). *P < 0.05, Wilcoxon signed-rank test.

Figure 3.. Whole transcriptome profiling of mock-treated, Cal/09- or Vic/11-infected monocytes reveals strain-specific alterations in the magnitude of the IFNAR response.

(A) Schematic of RNA-sequencing experimental design. Monocytes from three healthy blood bank donors were mock treated or infected with Cal/09 (MOI = 5) or Vic/11 (MOI = 0.2) followed by NK cell co-culture. At 7 or 24 HPI, monocytes were re-isolated by magnetic separation followed by RNA extraction, cDNA library preparation and paired-end illumina sequencing. (B) Principal component analysis (PCA) plot of the samples colored by treatment. (C) Heatmap showing genes differentially expressed by Cal/09 vs. Vic/11 infection. (D) KEGG pathways significantly enriched by each infection showing % overlap between strains. (E) Volcano plot of differentially expressed genes between Cal/09 and Vic/11 infected monocytes generated by DeSeq2, with the log2 fold change of each gene plotted against the total number of counts recorded for that gene. Differentially expressed genes with a p-value < 0.05 are highlighted in red. Triangles represent data points outside the graph area. (F) Differentially expressed genes were mapped to known protein-protein networks derived from the human STRING database. Individual gene are nodes in the network with scores assigned to them derived from a beta-uniform mixture model fitted to the unadjusted p-value distribution accounting for multiple testing. A subgraph (false discovery rate = 0.01) was generated with differential gene expression colored in red (upregulated; Cal/09 > Vic/11), green (downregulated; Cal/09 < Vic/11) or white (neutral). Shapes represent scores: rectangles are negative and circles are positive. Red cluster in bottom left: type 1 interferon-mediated signaling pathway elevated in Cal/09- vs. Vic/11-infected monocytes. Inset: IFNA1 mRNA counts within each condition.

Figure 4.. Cytokine profiles of Vic/11- and Cal/09-infected monocytes contribute to strain-specific NK cell IFN-γ responses.

(A) Cytokine concentrations assessed by Luminex®. Values displayed represent the fold-change in the mean fluorescence intensity between Cal/09- and Vic/11-infected conditions (MOI = 3). Cytokines elevated by 2.5-fold in supernatants harvested from Cal/09 or Vic/11-infected monocytes over the level in supernatants from mock-infected monocytes are plotted (n = 3). (B)-(F) Impact of cytokine receptor blocking on the NK cell IFN-γ response evaluated by pre-incubating NK cells for 1 hr with blocking antibodies specific to IL-12R (B), IL-15R (C), IL-18R (D), IFNGR1 (F) or with neutralizing antibody to IFN-α (F) followed by co-culture with infected monocytes. NK cell IFN-γ⁺ frequency compared with treatment with an isotype control antibody at 7 HPI (n = 3) or 24 HPI (n = 7–8) is shown. (G) IFN-α concentration following infection of monocytes with live or UV-inactivated Cal/09 or Vic/11 virus at an MOI of 3 (n = 6). *P < 0.05, Wilcoxon signed-rank test.

Figure 5.. Cytokine stimulation is insufficient to drive strain-specific NK cell anti-influenza IFN-γ production.

(A) Assessment of the impact of supernatants from infected cultures to the NK cell IFN-γ response to influenza strains. After infection (or mock-infection) for 24 hours, supernatants were then transferred to monocytes under each specified condition, followed by co-culture with autologous NK cells. Intracellular cytokine staining was used to assess the percentage IFN-γ⁺ NK cells by flow cytometry (n = 6). (B) Monocytes were cultured in direct contact with NK cells (contact), with monocytes seeded in dish and NK cells seeded in the trans-well (Trans-Well (1)), or monocytes seeded in the trans-well and NK cells seeded in the dish (Trans-Well (2)). (C) Percentage of IFN-γ⁺ NK cells after 6 hr co-culture with infected monocytes; Cal/09 (n = 6), Vic/11 (n = 5). (D) Cal/09- or Vic/11-infected monocytes were cultured alone or co-cultured with NK cells at a ratio of 1:1. Monocyte infection levels at 24 HPI assessed using an antibody specific to Flu-NP followed by intracellular flow cytometry, Cal/09 (n = 6), Vic/11 (n = 4). Mock-treated values were subtracted from infected measurements. *P < 0.05, Wilcoxon signed-rank test.

Figure 6.. Identification of influenza-mediated modulation of NK cell inhibitory and activating ligands by mass cytometry and GLMM analysis.

(A) Generalized linear mixed model (GLMM) to identify markers predictive of Cal/09 vs. Vic/11 infection from nine donors. Log-odds are logarithm of ratios of the probability that a cell is Vic/11-infected over the probability that a cell is Cal/09-infected at 24 HPI. An increase in the parameter coefficient corresponds to the strength of the classification power, with the 95% confidence interval from sampling error represented by line surrounding the point estimate. The confidence intervals are widened using Bonferroni’s method. The reported p-values are controlled using Benjamini-Hochberg’s false discovery rates. (B) PCA of individual cells colored by Cal/09 vs. Vic/11 infection at 24 HPI, with the vectors driving variance displayed. Quality check of variables that could potentially influence the ability to identify makers predictive of infection in monocyte – NK cell co-culture. (C) Cell count PCA of Cal/09 - vs. Vic/11-infected monocytes at 24 HPI. (D) PCA colored by donor of Cal/09 - vs. Vic/11-infected-infected monocytes at 24 HPI.

Figure 7.. Strain-specific downregulation of CD112 and CD54 is dependent on viral replication.

(A) Representative flow plots and histograms of CD112 expression on mock, Vic/11- or Cal/09-exposed monocytes and NHBE cells (MOI = 3). (B-C) The frequency of (B) CD112 (n = 9) or (C) CD54 (n = 8) expression on infected (Flu-NP⁺) monocytes assessed by fluorescence flow cytometry on Cal/09- or Vic/11-infected cells at an MOI of 3 at 24 HPI. (D-G) Representative histograms of CD112 (D) and CD54 (F) expression on mock-treated, live and UV-inactivated Cal/09- or Vic/11-infected monocytes. Summary plots of CD112⁺ (n = 9) (E) and CD54⁺ (n = 8) (G) expression after mock-treatment or infection with live or UV-inactivated Cal/09- or Vic/11-infected monocytes at MOI of 3. *P < 0.05, Wilcoxon signed-rank test.

Figure 8.. Cytokine-mediated and receptor-ligand interactions are necessary for the NK cell response to influenza-infected cells.

(A) NK cells were incubated for 1 hr with isotype control (Mouse IgG1, κ) or blocking antibodies specific to CD226 or LFA-1 followed by co-culture with autologous infected monocytes. At 24 HPI, intracellular cytokine staining was used to assess IFN-γ production compared to treatment with isotype control antibody (n = 6). (B) Schematic of fluorescently activated cell sorting to isolate uninfected monocytes expressing either low or high levels of CD54 and CD112. CD54^Low/CD112^Low and CD54^High/CD112^High monocytes were co-cultured with autologous NK cells for four hours in the presence of UV-treated supernatant harvested from mock, Cal/09- or Vic/11 monocytes (MOI = 3, 24 HPI) (n = 3 at 4 hr, n = 1 at 17 hr). (C) Compiled data from four donors from experiment described in (B). (D) Model of strain-specific NK cell recognition of influenza A infection. Cal/09 infection of monocytes elicits enhanced IFN-α secretion and neutralization of IFN-α dampens NK cell IFN-γ production. Cal/09-infection does not downregulate CD54 and CD112 to the same extent as Vic/11-infected monocytes or NHBE cells. *P < 0.05, Wilcoxon signed-rank test.