Natural killer cell-intrinsic type I IFN signaling controls Klebsiella pneumoniae growth during lung infection

Abstract

Klebsiella pneumoniae is a significant cause of nosocomial pneumonia and an alarming pathogen owing to the recent isolation of multidrug resistant strains. Understanding of immune responses orchestrating K. pneumoniae clearance by the host is of utmost importance. Here we show that type I interferon (IFN) signaling protects against lung infection with K. pneumoniae by launching bacterial growth-controlling interactions between alveolar macrophages and natural killer (NK) cells. Type I IFNs are important but disparate and incompletely understood regulators of defense against bacterial infections. Type I IFN receptor 1 (Ifnar1)-deficient mice infected with K. pneumoniae failed to activate NK cell-derived IFN-γ production. IFN-γ was required for bactericidal action and the production of the NK cell response-amplifying IL-12 and CXCL10 by alveolar macrophages. Bacterial clearance and NK cell IFN-γ were rescued in Ifnar1-deficient hosts by Ifnar1-proficient NK cells. Consistently, type I IFN signaling in myeloid cells including alveolar macrophages, monocytes and neutrophils was dispensable for host defense and IFN-γ activation. The failure of Ifnar1-deficient hosts to initiate a defense-promoting crosstalk between alveolar macrophages and NK cell was circumvented by administration of exogenous IFN-γ which restored endogenous IFN-γ production and restricted bacterial growth. These data identify NK cell-intrinsic type I IFN signaling as essential driver of K. pneumoniae clearance, and reveal specific targets for future therapeutic exploitations.

Fig 1. Type I IFN signaling protects against K. pneumoniae lung infection by controlling bacterial growth and lung pathology.

(A) WT and Ifnar1^-/- mice (n = 12 per genotype) were infected intranasally (5 x 10⁴ CFU of K. pneumoniae), and survival was monitored for 10 days. Kaplan-Meier survival curves are shown. Statistical evaluation: Log-rank (Mantel-Cox) test. **, P < 0.01. (B) Weight changes during the first 3 days following infection. Data are represented as mean ± SEM. Statistical evaluation: unpaired Student’s t test. ***, P < 0.001. (C, D) WT and Ifnar1^-/- mice were infected as in (A) and bacterial loads were determined in lungs 12, 24 and 48 h p. i. (C), and in spleens and livers 48 h p. i. (D). Bacterial load is presented as CFU per g of analyzed organ per infected animal. Dot plots: horizontal bars represent median. Statistical evaluation: Mann-Whitney test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. (E, F) H&E-stained sections of lungs of infected WT and Ifnar1^-/- animals at 48 (E) and 72 h (F) p. i., together with quantification of airway inflammation, intralesional bacterial burden, neutrophilic infiltration and the combined histopathology analysis (total histopathology score) (n = 7, 6, 9 and 10 mice for WT 48 h p. i., Ifnar1^-/- 48 h p. i., WT 72 h p. i., Ifnar1^-/- 72 h p. i., respectively). Insets: Arrows and N indicate neutrophils. Note an increased airway inflammation, intralesional bacterial burden and neutrophilic infiltration in lungs from Ifnar1^-/- animals. Histopathology scores were determined by blinded scoring; error bars, mean ± SEM. Statistical evaluation: Mann-Whitney test. *, P < 0.05; ns, not significant.

Fig 2. Induction of type I IFN signaling in macrophages by K. pneumoniae is dependent on Irf3 and Tlr4, and the bacterial capsule polysaccharide (CPS) and LPS O-polysaccharide.

(A) Mouse alveolar macrophages isolated from K. pneumoniae-infected (intranasal, 5 x 10⁴ CFU) or PBS-treated WT mice were analyzed for expression of Ifnb, Isg15 and Tnf using qPCR 24 h p.i. (n = 5, PBS; n = 6, infection). (B) BMDMs from WT, Irf3^-/- and Ifnar1^-/- mice were left untreated or infected for indicated time points with K. pneumoniae (MOI = 70), and mRNA levels of Mx1, Ifit1, Isg15 and Tnf were determined by qPCR. Error bars, mean ± SEM (n > 3). (C, D) WT and Tlr4^-/- BMDMs were infected as in (B) for indicated time points, or left untreated. Phospho-TBK1 (p-TBK1), phospho-IRF3 (p-IRF3) and tubulin (loading control) were detected in whole cell extracts by Western blotting (E), and Ifnb, Mx1, Ifit1 and Isg15 mRNA levels were quantitated by qPCR (D). (E) BMDMs were infected as in (B) with a cps, O-polysaccharide, and double cps-O-polysaccharide K. pneumoniae mutants for indicated time points, or left untreated, and type I IFN levels in the supernatants were quantitated using bioassays. Statistical evaluation in (A) and (E): unpaired Student’s t test; error bars, mean ± SEM (n > 3); *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Fig 3. Type I IFN signaling is indispensable for IFN-γ, IL-12 and CXCL10 induction in K. pneumoniae-infected lungs.

WT and Ifnar1^-/- mice were infected intranasally (5 x 10⁴ CFU of K. pneumoniae) for 12 (A, B) or 48 (C) h, or treated with PBS, and gene expression was determined by qPCR normalized to Hprt. (A) Mx1, Ifit1 and Isg15 mRNA levels in lungs. (B, C) Ifng, Il12b, Cxcl10, Tnf and Il10 mRNA levels in lungs. Statistical evaluation: unpaired Student’s t test; error bars, mean ± SEM (n > 3); *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Fig 4. Type I IFN signaling induces NK cell accumulation and activation in lungs during K. pneumoniae infection.

WT and Ifnar1^-/- mice were infected intranasally (5 x 10⁴ CFU of K. pneumoniae) for 12 h or treated with PBS and immune cell subsets in lungs were analyzed by flow cytometry. Cells were subgated for CD45⁺ cells. (A) Alveolar macrophage, neutrophil and inflammatory monocyte populations were detected as SiglecF⁺CD11c^high, Cd11b⁺Ly6G⁺Ly6C^med, and CD11b⁺Ly6G^-Ly6C^high, respectively. Populations are presented as total cells per lung calculated from percentages of individual subsets (shown in S4 Fig). (B, C) NK cells, detected as CD3^-NK1.1⁺, and IFN-γ⁺ NK cells are shown. Representative plots of NK cells (B, left panels) and IFN-γ⁺ NK cells (C, left panels) are shown. Numbers indicate percentages in the outlined area of live, CD45⁺ cells. NK cell population in dot plots is shown both as percent of CD45⁺ cells as well as total cells per lung (B, right panels). IFN-γ⁺ cells are shown both as % of CD3^-NK1.1⁺ cells and as total cells per lung (C, right panels). (D-G) CD4 T cells detected as CD3⁺CD4⁺ (D), IFN-γ-producing CD4 T cells detected as IFN-γ⁺ CD3⁺CD4⁺ (E), CD8 T cells detected as CD3⁺CD8⁺ (F), and IFN-γ-producing CD8 T cells detected as IFN-γ⁺ CD3⁺CD8⁺ (G) are shown in dot plots. CD4 and CD8 T cells are shown as percent of CD45⁺ cells (D and F, left panels) and as total cells per lung (D and F, right panels). IFN-γ⁺ CD4 and IFN-γ⁺ CD8 T cells are shown both as % of CD3⁺CD8⁺ cells (E and G, left panels) and as total cells per lung (E and G, right panels). Statistical evaluation: unpaired Student’s t test; error bars, mean ± SEM (n > 3); *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Fig 5. Priming by IFN-γ promotes macrophage activation against K. pneumoniae.

(A) Alveolar macrophages from WT and Ifnar1^-/- mice were pretreated or not with 5 ng/ml mouse IFN-γ for 5 h, followed by infection (MOI = 70) for 1 h and RNA isolation. mRNA levels of Il12b, Cxcl10, Tnf and Il1b were quantitated by qPCR and normalized to Hprt. (B,C) Macrophages primed or unprimed with IFN-γ (10 ng/ml) for 2 h were infected (MOI = 70) for 90 min (including gentamicin treatment starting at 30 min after infection) (B) or 30 min without gentamicin treatment (C), and bacterial loads were determined as CFU per ml. (D-F) WT and Irf3^-/- mice were infected intranasally (5 x 10⁴ CFU of K. pneumoniae) for 48 h (n = 11 per genotype). Expression of indicated genes (D,E) and bacterial loads (F) in lungs were determined using qPCR and CFU assays, respectively. (G, H) WT mice received intranasally IFNβ (30 μl, 30,000 U) or PBS (n = 4 per treatment). Animals were euthanized 6 h following treatment. RNA was isolated from lungs and analyzed for expression of Mx1, Isg15 and Ifit1 (G) as well as Ifng, Il12b and Cxcl10 (H). Statistical evaluation in (A), (D), (E), (G) and (H): unpaired Student’s t test; error bars, mean ± SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. Statistical evaluation in (B), (C) and (F): Mann-Whitney test; horizontal bars represent median; ***, P < 0.001; ns, not significant.

Fig 6. Protective responses against K. pneumoniae infection develop independently of type I IFN signaling in alveolar macrophages.

(A) Ifnar1^fl/fl-CD11cCre and Ifnar1^fl/fl mice (n = 11 and 10, respectively) were infected intranasally (5 x 10⁴ CFU of K. pneumoniae), and survival was monitored for 10 days. Kaplan-Meier survival curves are shown. Statistical evaluation: Log-rank (Mantel-Cox) test; ns, not significant. (B, C) Ifnar1^fl/fl-CD11cCre and Ifnar1^fl/fl mice (n = 6 per genotype) were infected intranasally (5 x 10⁴ CFU of K. pneumoniae) or treated with PBS (n = 2 and 3, respectively) for 12 h. Expression of Mx1 and Ifit1 (B), and Ifng, Il12b, Cxcl10, Tnf, Il10 and Cxcl1 (B) in lungs was determined by qPCR (normalization to Hprt). Statistical evaluation: unpaired Student’s t test; error bars, mean ± SEM (n > 3); *, P < 0.05; **, P < 0.01; ns, not significant. (D) Ifnar1^fl/fl-CD11cCre and Ifnar1^fl/fl mice (n = 6 per genotype) were infected intranasally (5 x 10⁴ CFU of K. pneumoniae) for 12 h or treated with PBS and immune cell subsets in lungs were analyzed by flow cytometry as in Fig 4. Populations are presented as total cells per lung calculated from percentages of individual immune cell subsets (shown in S5 Fig). Statistical evaluation: unpaired Student’s t test; error bars, mean ± SEM; ns, not significant.

Fig 7. Transfer of WT NK cells or administration of IFN-γ restore control of K. pneumoniae growth in Ifnar1^-/- mice.

(A) Ifnar1^-/- mice were treated with PBS, infected intranasally (5 x 10⁴ CFU of K. pneumoniae) (K.p.), or given 1 x 10⁶ WT NK cells and infected intranasally (5 x 10⁴ CFU of K. pneumoniae) (K.p. + WT NK). Lungs were analyzed 24 h p.i. by flow cytometry for IFN-γ-producing endogenous Ifnar1^-/- NK cells (dot plot groups 1–3) and exogenous WT NK cells (dot plot group 4) shown as percent of IFN-γ⁺ in CD3^-NK1.1⁺ cells. Statistical evaluation: dots represent individual mice; error bars, mean ± SEM; unpaired Student’s t test; *, P < 0.05; **, P < 0.01; ns, not significant. (B) WT mice were treated with PBS, infected intranasally (5 x 10⁴ CFU of K. pneumoniae) (K.p.), or given 1 x 10⁶ WT NK cells and infected intranasally (5 x 10⁴ CFU of K. pneumoniae) (K.p. + WT NK). Lungs were analyzed 24 h p.i. by flow cytometry for IFN-γ-producing endogenous WT NK cells (dot plot groups 1–3) and exogenous WT NK cells (dot plot group 4) shown as percent of IFN-γ⁺ in CD3^-NK1.1⁺ cells. Statistical evaluation: dots represent individual mice; unpaired Student’s t test; **, P < 0.01; ns, not significant. (C, D) Bacterial loads in lungs (C) and spleen (D) of infected WT and Ifnar1^-/- mice which received WT NK cells or were mock-treated. Data show CFU per g of lung per infected animal. Dots in dot plots represent individual mice. Statistical evaluation: one-way ANOVA with multiple comparisons; *, P < 0.05; ***, P < 0.001; ns, not significant. (E, F) Ifnar1^-/- mice infected intranasally (5 x 10⁴ CFU of K. pneumoniae) and given IFN-γ or PBS (mock) at the time of infection. Mice were euthanized 24 h p.i. and mRNA expression (Ifng, Il12, Cxcl10, Tnf) (E) and bacterial loads (F) in lungs were determined by qPCR and CFU assays, respectively. Statistical evaluation in (E) (n = 5, mock; n = 7, IFN-γ): unpaired Student’s t test; error bars, mean ± SEM; **, P < 0.01; ***, P < 0.001; ns, not significant. Statistical evaluation (F) (n = 6, mock; n = 7, IFN-γ): Mann-Whitney test; *, P < 0.05.