Natural Killer Cell Sensing of Infected Cells Compensates for MyD88 Deficiency but Not IFN-I Activity in Resistance to Mouse Cytomegalovirus

Abstract

In mice, plasmacytoid dendritic cells (pDC) and natural killer (NK) cells both contribute to resistance to systemic infections with herpes viruses including mouse Cytomegalovirus (MCMV). pDCs are the major source of type I IFN (IFN-I) during MCMV infection. This response requires pDC-intrinsic MyD88-dependent signaling by Toll-Like Receptors 7 and 9. Provided that they express appropriate recognition receptors such as Ly49H, NK cells can directly sense and kill MCMV-infected cells. The loss of any one of these responses increases susceptibility to infection. However, the relative importance of these antiviral immune responses and how they are related remain unclear. In humans, while IFN-I responses are essential, MyD88 is dispensable for antiviral immunity. Hence, a higher redundancy has been proposed in the mechanisms promoting protective immune responses against systemic infections by herpes viruses during natural infections in humans. It has been assumed, but not proven, that mice fail to mount protective MyD88-independent IFN-I responses. In humans, the mechanism that compensates MyD88 deficiency has not been elucidated. To address these issues, we compared resistance to MCMV infection and immune responses between mouse strains deficient for MyD88, the IFN-I receptor and/or Ly49H. We show that selective depletion of pDC or genetic deficiencies for MyD88 or TLR9 drastically decreased production of IFN-I, but not the protective antiviral responses. Moreover, MyD88, but not IFN-I receptor, deficiency could largely be compensated by Ly49H-mediated antiviral NK cell responses. Thus, contrary to the current dogma but consistent with the situation in humans, we conclude that, in mice, in our experimental settings, MyD88 is redundant for IFN-I responses and overall defense against a systemic herpes virus infection. Moreover, we identified direct NK cell sensing of infected cells as one mechanism able to compensate for MyD88 deficiency in mice. Similar mechanisms likely contribute to protect MyD88- or IRAK4-deficient patients from viral infections.

Fig 1. pDC or MyD88 deficiency impairs IFN-I production but not splenic IFN-I responses and virus control.

Control (Rat IgG), pDC-depleted (αBST2) or untreated BALB/c mice and untreated BALB/c, BALB/c MyD88^-/-, BALB/c IFNAR^-/- and BALB/c TLR9^-/- mice were infected with 2.5 x 10³ pfu MCMV or left uninfected (NI). (A) IFN-α serum titers were measured at d1.5 post-infection by ELISA. Results (mean±SEM) are shown from one experiment representative of 3 independent ones. (B) RT-PCR analysis of the expression of selected genes in the spleen at d1.5 or 3 post-infection. Data are normalized to mean expression of d1.5 BALB/c mice (100% reference level). Results (mean±SEM) are shown from 2 pooled independent experiments, each with 2 to 3 mice per group. (C) Serum IFN-α levels were measured at d1.5 post-infection by ELISA in BALB/c-Ly49H⁺, BALB/c-Ly49H⁺ MyD88^-/-, BALB/c and BALB/c MyD88^-/- mice. Results (mean±SEM) are shown from one experiment representative of 3 independent ones, each with 3 mice per group. BLD: below limit of detection. (D-E) Microarray analyses were performed on total mRNA extracted from spleen at d0, 1.5, 2, 3 and 6 post-infection. Heatmaps show the relative expression value for IFN-I/III genes (D) and for 25 ISG (E). The expression pattern of ISG is examined globally in S2E–S2G Fig. Results shown are from 2 pooled independent experiments, each with 1 to 3 mice per group.

Fig 2. IFN-I responses are essential but MyD88 and Ly49H partly redundant to protect against MCMV infection.

(A) Survival of mice at d21 post-infection (Y-axis) as a function of the doses of MCMV inoculum (X-axis). Indicated mice were infected with between 1.6x10³ and 7x10⁴ PFU of MCMV, with overlaps for several doses between strains of mice, and their survival was followed for 21d. Data is shown as percentage of survival. In total, 8 different doses were tested in 9 different experiments in order to determine the susceptibility of each mouse strain, with 2 to 4 mouse strains simultaneously studied in each experiment. Data was derived from the following numbers of mice and tested doses of MCMV inoculum: BALB/c-Ly49H⁺: 24 mice for 4 doses; BALB/c-Ly49H⁺ MyD88^-/-: 37 mice for 6 doses; BALB/c-Ly49H⁺ IFNAR^-/-: 8 mice for 2 doses; BALB/c: 43 mice for 5 doses; BALB/c MyD88^-/-: 15 mice for 4 doses; BALB/c IFNAR^-/-: 6 mice for 2 doses. The median lethal dose that causes 50% of lethality in each mouse strain (LD₅₀) was calculated from the graph as the dose on the X-axis that corresponds to 50% mortality on the Y-axis as represented by the horizontal dotted line labeled “LD₅₀ calculation”. LD₅₀ BALB/c-Ly49H⁺: ≥10⁵ pfu/mouse; LD₅₀ BALB/c-Ly49H⁺ MyD88^-/-: 3.3x10⁴ pfu; LD₅₀ BALB/c-Ly49H⁺ IFNAR^-/-: 2x10³ pfu; LD₅₀ BALB/c: 2.2x10⁴ pfu; LD₅₀ BALB/c MyD88^-/-: 8x10³ pfu; LD₅₀ BALB/c IFNAR^-/-: 2x10³ pfu. (B) Splenic viral titers were measured at d1.5, 3 and 6 post-infection. Data (mean±SEM) are shown from 2 to 4 pooled independent experiments, each with 2 to 3 mice per group.

Fig 3. MyD88 deficiency does not completely abrogate NK cell-mediated protection.

(A-D) Ly49H⁺ NK cell activation in BALB/c-Ly49H⁺, BALB/c-Ly49H⁺ MyD88^-/- and BALB/c-Ly49H⁺ IFNAR^-/- mice at d0, 3 and 6 post infection. Proliferation was assessed by Ki67 expression (A). Antiviral effector functions were assessed by intracellular staining for IFN-γ (B) and Granzyme B (C-D). (E) Impact of NK cell depletion on viral replication in the spleen of BALB/c-Ly49H⁺, BALB/c-Ly49H⁺ MyD88^-/- and BALB/c-Ly49H⁺ IFNAR^-/- mice. Mice were depleted of NK cells and splenic viral titers were measured 5 or 6d post infection. (F-I) Ly49H^- NK cell activation in BALB/c-Ly49H⁺, BALB/c-Ly49H⁺ MyD88, BALB/c-Ly49H⁺ IFNAR^-/-, BALB/c, BALB/c MyD88^-/- and BALB/c IFNAR^-/- mice. Data (mean±SEM) are shown from 3 pooled independent experiments each with 3 mice per group.

Fig 4. Impact of MyD88 or Ly49H deficiencies on antiviral CD8 T cell responses.

(A-C) Analysis of antiviral CD8 T cell effector functions. (A) In vivo cytotoxicity, ex vivo IFN-γ production and Granzyme B expression by splenic CD8 T cells from BALB/c-Ly49H⁺, BALB/c-Ly49H⁺ MyD88^-/-, BALB/c and BALB/c MyD88^-/- mice at d0 and 6 post infection. Data (mean ± SEM) are shown from 3 pooled independent experiments each with 2 or 3 mice per group. (B-C) Impact of CD8 T cell depletion on disease in BALB/c and BALB/c MyD88^-/- mice. Mice were depleted of CD8 T cells and infected with 2.5x10³ pfu MCMV (B) or with 8x10³ pfu MCMV (C). Splenic viral titers were measured at d6 post infection (B) or mortality was monitored daily (C). For B, data (mean±SEM) are represented from 2 pooled independent experiments each with 2 to 4 mice per group. For C, data show the percent survival from 2 pooled independent experiments each with 3 to 5 mice per group, n represents the total number of mice per group. (D) Enumeration of total (left panel) and anti-IE-1 (right panel) CD8 T cells in the spleens of d5 MCMV-infected mice. Results (mean±SEM) are shown from 4 pooled independent experiments each with 3 mice per group. (E) Immunohistological analysis of tissue damage to the spleen in d4 MCMV infected mice. Mice were infected with 10⁴ pfu MCMV. Spleen were harvested at d4 and stained to evaluate the integrity of the T cell zone as assessed by examining its marginal zone boundary (CD169) and its T cell (CD3) and B cell (B220) zones. Results are shown for one representative mouse per experimental group from 3 independent experiments each with 3 mice per group.

Fig 5. Model of redundancies and complementarities between molecular sensors and cell types for mounting the IFN-I, IFN-γ and cytotoxic cellular immune responses which are necessary for control of MCMV infection.

Based on our own data as well as on many other studies published previously, we propose a model whereby IFN-I, IFN-γ and cytotoxic responses are all necessary for immune defenses against MCMV infection (red lines and text), but access to these functions can be promoted by a number of partly redundant and/or complementary pathways (arrows converging towards a number). IFN-I, IFN-γ and cytotoxic cellular immune responses are critical for control of viral replication, prevention of excessive tissue damage and overall resistance of the host in terms of morbidity and mortality. Several cell types and pathways can lead to IFN-I production. Early after intra-peritoneal injection, MCMV infects stromal cells and other cell types, inducing an IFN-I production by those cells independently of MyD88 and TLRs. In addition, viral particles or material derived from infected cells can be engulfed by cDC and pDC to promote their production of high levels of IFN-I or other cytokines upon triggering of the TLR7/9-to-MyD88 signaling cascade. In our experimental settings, MyD88/TLR9 responses of DC and MyD88-independent responses of infected cells are redundant for the induction of strong IFN-I responses, even though only very low to undetectable levels of IFN-I are produced in the absence of MyD88 responses (❶). In parallel, NK cells and CD8 T cells are able to produce IFN-γ and to specifically recognize and kill MCMV infected cells, via the activation receptor Ly49H or via their TCR, respectively. Cell-mediated immune control of viral replication can thus be performed by both NK and CD8 T cells which can largely compensate one another for this function (❷). However, the antiviral functions of NK and CD8 T cells are not strictly redundant but partly complementary, since, for example, the absence of CD8 T cell responses might increase the risk of selection of viral mutants able to escape NK cell control. MyD88 responses can promote the activation of NK and CD8 T cells via TLR-dependent activation of myeloid cells and/or through cell-intrinsic effects. However, MyD88 responses are not necessary for this function in mice expressing NK cell activation receptors able to directly sense MCMV-infected cells. In these mice, in addition to IFN-I production, MyD88-independent responses from infected cells might provide the other beneficial inflammatory signals necessary for this function (❸). Functional redundancies might also exist between MyD88 responses and NK cell activity for prevention of excessive damage to lymphoid tissues (❹), in order to preserve their micro-anatomical niches supporting the proliferation and survival of CD8 T cells.