Interleukin-18 improves the early defence system against influenza virus infection by augmenting natural killer cell-mediated cytotoxicity

Influenza virus attacks a large proportion of the population every year and can cause death, especially among older people, infants and immunocompromised individuals (Brody & Brock, 1985; Glezen et al., 1980). The immunocompetent cell system with a network of cytokines and chemokines induces protective immunity against influenza virus infection (Ryan et al., 2002; Van Reeth et al., 2002). Interleukin (IL)-18, previously considered to be an interferon-gamma (IFN-γ)-inducing factor, is synthesized by activated macrophages and shows similar properties to IL-1 in biochemical functions and to IL-12 in biological functions (Dinarello, 1999; Okamura et al., 1998). This cytokine plays a critical role in the development of protective immunity against intracellular pathogens including Mycobacterium tuberculosis, Yersinia enterocolitica, Cryptococcus neoformans and herpes simplex virus (Bohn et al., 1998; Fujioka et al., 1999; Kawakami et al., 1997; Sugawara et al., 1999). The protective role of IL-18 is attributable to its ability to induce IFN-γ production, to activate natural killer (NK) cells and to proliferate activated T lymphocytes (Takeda et al., 1998). We have reported that IL-18 plays a key role in activating microglial functions by inducing neuronal IFN-γ in the brain parenchyma following neurovirulent influenza A virus infection in the olfactory bulb (Mori et al., 2001). However, knowledge is scarce about the relationship between the host defence system in the respiratory tract and the role of endogenous IL-18 during influenza virus infection. IL-18 mRNA has been found to be constitutively expressed within the lung, mostly localized in the respiratory epithelial cells (Cameron et al., 1999). Influenza virus principally causes surface infection that is restricted to mucosal cells of the respiratory tract. Therefore, it is interesting to investigate whether IL-18 is a critical mediator for establishment of the primary defence system against influenza virus infection.

In the present study, we infected IL-18-gene disrupted (IL-18^-/-) mice with influenza A virus intranasally and investigated the host defence mechanism, particularly focusing on the local responses in the respiratory tract. The response of the wild-type C57BL/6 mice to influenza virus infection was used as a control.

Virus.
The mouse-adapted strain of human influenza virus A/PR/8/34 (H1N1) was propagated routinely by allantoic inoculation of 10-day-old embryonated chicken eggs with seed virus diluted 1 : 10^-4. Virus titre was assayed by plaque titration on MadineDarby canine kidney cell monolayers, as described elsewhere (Mori et al., 1995).

Mice.
IL-18^-/- mice, with the genetic background of C57BL/6 mice, were generated by the gene targeting method (Takeda et al., 1998). The mice were bred and maintained in the Animal Laboratory of Fukui Medical University under specific pathogen-free conditions. The concentration of IL-18 molecules in serum from IL-18^-/- mice, measured by ELISA, was below the detectable level (cf. 1000±20 pg ml^-1 in normal C57BL/6 mice). This indicated that the IL-18 gene mutation leads to a lack of IL-18 production. Six-week-old IL-18^-/- mice and age-matched C57BL/6 mice (Clea Japan) were used for the experiments.

Experimental infections.
Mice were mildly anaesthetized by intraperitoneal administration of pentobarbital sodium (0·025 mg g^-1 body mass) and inoculated in the right nostril with influenza A virus in 20 µl sterile PBS. To avoid laboratory contamination, all virus-infected mice were housed in negatively pressurized isolators equipped with a ventilation system through a high-efficiency particulate air filter (AH model; Nihon-Ika). This work was approved by the Committee of Institutional Animal Care and Use in Fukui Medical University.

Preparation of single-cell suspensions from the lung parenchyma.
Mice were anaesthetized and the lung was flushed in situ with 20 ml sterile PBS via cannulation of the heart to remove the intravascular blood pool. Minced lung tissues were incubated at 37 °C for 60 min on a rocker with 200 µg collagenase D ml^-1 and 40 µg DNase I ml^-1 (both from Roche Molecular Biochemicals). Subsequently, the enzyme-digested lung tissue was passed through a stainless steel mesh. Single-cell suspensions were collected by density-gradient centrifugation with lymphocyte separation solution (Antibody Institute).

Identification of lung parenchyma cells.
Single-cell suspensions from the lung parenchyma were identified by flow cytometry (EPCS XL, Beckman Coulter) using monoclonal antibodies for CD4, CD8, CD16/32 and neutrophil (Caltag).

Assay of cytokine and nitric oxide production in the BAL fluids.
The amounts of cytokines such as IFN-γ, IL-12 and IL-4 were assayed by using a mouse cytokine detection ELISA kit (BioSource) as described previously (Liu et al., 2003). The minimum detectable concentrations of IFN-γ, IL-12 and IL-4 were 1 pg ml^-1, 2 pg ml^-1and 5 pg ml^-1, respectively. The level of nitric oxide was measured with a Nitric Oxide Assay kit (Calbiochem-Novabiochem), in accordance with the manufacturer's instructions.

Assay of NK cell and cytotoxic T lymphocyte (CTL) activities in the lung parenchyma.
A cytotoxicity assay was performed according to the protocol previously described (Liu et al., 2001). Lung parenchyma cells and NK-sensitive Yac-1 target cells were mixed and incubated at 37 °C in a 5 % CO₂ atmosphere for 4 h. Specific lysis of target cells was determined by a lactate dehydrogenase release assay (Decker & Lohmann-Matthes, 1988) using a Cytotoxic Detection kit (Roche). Data were expressed as the percentage of specific release: 100x[(target with effector - effector spontaneous) - target spontaneous]/[target maximum - target spontaneous]. For the assay of specific CTL activity, target cells were prepared using mouse lymphoma EL-4 cells infected with influenza A virus at an input m.o.i. of 10 p.f.u. per cell.

Antibody assay.
Virus-specific immunoglobulins (Igs) were measured with an ELISA Ig Quantitative kit (Bethyl Laboratories) as described previously (Dong et al., 2003). Bronchoalveolar lavage (BAL) fluids and sera were collected and assayed for a T helper type 2 (Th2)-related antibody, IgG1, and for a Th1-related antibody, IgG2a.

Statistical significance.
The two-tailed MannWhitney U-test and Student's t-test were used to determine whether a significant difference (P<0·05) existed between IL-18^-/- and control C57BL/6 mice.

Susceptibility of IL-18^-/- mice to influenza virus infection
To elucidate the role of IL-18 molecules in host defence mechanisms against influenza virus infection, mice were inoculated intranasally with 10³ p.f.u. influenza A/PR/8/34 virus per mouse and the time course of mortality was investigated. Seventeen per cent of virus-infected IL-18^-/- mice died within the observation period of 15 days, while all wild-type C57BL/6 mice survived the infection (Fig. 1a). When the inoculum dose was increased to 10⁴ p.f.u. per mouse, the overall survival rate was 33 % for IL-18^-/- and 50 % for wild-type mice (Fig. 1b). The LD₅₀ of influenza virus was calculated as 10^3·7 p.f.u. for IL-18^-/- mice and 10⁴ p.f.u. for wild-type mice. The difference in the mortality between the IL-18^-/- and the wild-type mice was at no point statistically significant. Virus growth in the lungs of IL-18^-/- mice reached the maximum more quickly, peaking at day 3, with a significantly higher virus titre than that of wild-type mice (Fig. 2). The progeny virus was eventually cleared from the lungs of both IL-18^-/- and wild-type mice by day 12 after virus inoculation.

(18K): Fig. 1. Survival profiles of IL-18-deficient (•) and wild-type C57BL/6 () mice after intranasal infection with influenza A/PR/8/34 virus at an inoculum dose of 103 (a) and 104 (b) p.f.u. per mouse (n=12).

(17K): Fig. 2. Pulmonary virus growth in IL-18-deficient (filled bar) and wild-type C57BL/6 (open bar) mice after intranasal inoculation with influenza A virus at an inoculum dose of 103 p.f.u. per mouse. Data are means±SD of results for each group of six mice tested. *, Significant difference compared with corresponding wild-type C57BL/6 mice (P<0·01).

Inflammation in the lungs of IL-18^-/- mice
Single-cell suspensions were collected from the lungs and identified using specific antibodies for cell markers. A large number of cells infiltrated the lung parenchyma of IL-18^-/- mice during the early phase of infection (Fig. 3). The cell population in the lungs consisted mostly of neutrophils (Table 1). Large quantities of nitric oxide could be detected in BAL fluids immediately after infection (Fig. 4a). In sera, larger amounts of nitric oxide appeared on day 2 and quickly vanished on day 3 (Fig. 4b), giving a possible reason for severe respiratory damage. The dominant inflammatory reaction in IL-18^-/- mice at this time point after infection was correlated with a large amount of virus loading in the lung parenchyma (Fig. 2). It should be noted that the number of T lymphocytes positive for CD4 or CD8 did not differ between the two strains of mice (Table 1).

(13K): Fig. 3. Cellular infiltration in the lung of IL-18-deficient (filled bar) and wild-type C57BL/6 (open bar) mice after intranasal infection with influenza A virus at an inoculum dose of 103 p.f.u. per mouse. Data are means±SD of results for each group of six mice tested. *, Significant difference compared with corresponding wild-type C57BL/6 mice (P<0·01).

Table 1. Distribution of cell populations in the lung of IL-18-deficient (IL-18-/-) and wild-type C57BL/6 mice Mice were infected intranasally with influenza virus at an inoculum dose of 103 p.f.u. per mouse. Lung cells were collected on day 2 after infection and analysed by flow cytometry. Data are means±SD of results for each group of six mice tested.

(13K): Fig. 4. Induction of nitric oxide in (a) bronchoalveolar lavage fluids and (b) sera of IL-18-deficient (filled bar) and wild-type C57BL/6 (open bar) mice after influenza A virus infection. Data are means±SD of results for each group of six mice tested.

Effect of IL-18 molecules on local cytokine production
Mice were infected with influenza virus intranasally and BAL fluids were collected and assayed for IFN-γ, IL-12 and IL-4 titres. IFN-γ production in wild-type C57BL/6 mice was increased for the first 3 days of infection, while in IL-18^-/- mice only a slight increase in titre could be detected (Fig. 5). In contrast to IFN-γ, IL-18^-/- mice produced larger amounts of IL-12 upon infection compared with wild-type mice. The level of IL-4, a Th2 cytokine, was not detectable until 9 days after infection, at which time a very low titre was measured in both strains of mice.

(17K): Fig. 5. Cytokine production in the bronchoalveolar lavage fluid of IL-18-deficient (filled bar) and wild-type C57BL/6 (open bar) mice after intranasal inoculation with influenza A virus at an inoculum dose of 103 p.f.u. per mouse. At intervals after infection, the bronchoalveolar lavage fluids were collected and assayed for (a) IFN-γ, (b) IL-12 and (c) IL-4. Data are means±SD of results for each group of six mice tested. *, Significant difference compared with corresponding wild-type C57BL/6 mice (P<0·01).

NK cell activity of lung parenchyma cells
During an early stage of infection, before the appearance of the specific immune responses, NK cell activity is the major factor that contributes to a rapid termination of virus infection. As anticipated, lung parenchyma cells from IL-18^-/- mice showed a lower level of activity of NK cell-mediated cytolysis than those from wild-type mice (Fig. 6). This finding indicated that IL-18 is essential for full development of NK cell activity during the early days of infection.

(20K): Fig. 6. Natural killer cell activity of the lung cells from influenza A virus-infected IL-18-deficient (•) and wild-type C57BL/6 () mice at the effector to target cell ratio of 50 : 1. Data are means±SD of results for each group of six mice tested. *, Significant difference compared with corresponding wild-type C57BL/6 mice (P<0·01).

Specific humoral and cellular immunity in IL-18^-/- mice
Specific Th2-related IgG1 and Th1-related IgG2a antibody was produced in both BAL fluids and sera of IL-18^-/- mice upon influenza virus infection (Table 2). The antibody titres were equivalent to those of wild-type mice. In addition, the level of virus-specific CTL activity induced in the lung of IL-18^-/- mice was comparable with the level in wild-type mice (Table 2). These results suggested that the IL-18 molecule has no appreciable influence on the induction of ordinary immune responses.

Table 2. Influenza virus-specific local immune responses in IL-18-deficient (IL-18-/-) and wild-type C57BL/6 mice Mice were infected intranasally with influenza A virus at an inoculum dose of 103 p.f.u. per mouse. The bronchoalveolar lavage (BAL) fluids and sera were collected 3 weeks after infection and assayed for influenza virus-specific antibody by ELISA. Antibody titres (log ng ml-1) of IL-18-/- and C57BL/6 mice before virus infection were less than 0·6 in BAL and less than 1·3 in serum. Influenza virus-specific cytotoxic T lymphocyte (CTL) activity on day 9 after infection was determined. Data are means±SD of results for each group of six mice tested.

IL-18 was originally described as a factor for IFN-γ production of T lymphocytes in the presence of IL-12. Recent studies have demonstrated that IL-18 shows many other biological activities, including proliferation of T lymphocytes and NK cells, stimulation of their cytotoxic activity and enhancement of a Th1-mediated immune response (Akira, 2000; Dinarello, 1999; Kohno et al., 1997). On the other hand, IL-18 acts as a proinflammatory cytokine and increases the severity of diseases with lethal endotoxaemia (Dinarello, 2000; Lauw et al., 1999; Netea et al., 2000). Importantly, IL-18 elicits antiviral activity in the acute phase of infection. In the case of vaccinia virus infection, IL-18 is involved in various host defence mechanisms, including NK cells and CTLs (Tanaka-Kataoka et al., 1999). In fact, virus-induced IL-18 and IFN-γ enhance Fas ligand expression on NK cells (Tsutsui et al., 1996) and Fas molecules on virus-infected cells (Takizawa et al., 1993).

The results presented in this paper show that, during the early stage of influenza virus infection, just before the appearance of specific immune responses, the reduced NK activity of lung parenchyma cells in IL-18^-/- mice was correlated with profound virus growth in the lung. Thus, IL-18 plays an important role in the early antiviral host response by up-regulation of the NK cell-mediated killing activity of virus-infected cells. A similar early antibacterial action of IL-18 has also been found during Streptococcus pneumoniae infection (Lauw et al., 2002).

IFN has long been recognized as an essential part of the innate cytokine response to virus infection (Hennet et al., 1992), showing immunomodulatory activity as well as direct antiviral activity. During the process of IFN-γ production, IL-18 acts in synergy with IL-12 (Kohno et al., 1997). It is interesting to note that, despite the high titre of IL-12 induced in IL-18^-/- mice, production of IFN-γ was still at a low level (Fig. 5). This finding implies that IL-12 alone, in the absence of IL-18, is insufficient for activation and development of IFN-γ production in vivo and that functions operated by IL-18 molecules cannot fully be compensated by IL-12. The major cell population responsible for IFN-γ production by IL-18 stimulation is that of T lymphocytes and NK cells (Hunter et al., 1997). The lower levels of production of IFN-γ in IL-18^-/- mice during the early days of infection might be due to a reduction in the number of IFN-γ-expressing NK cells (Pien et al., 2000), although the total number of NK cells in IL-18^-/- mice was slightly greater than in wild-type mice (Table 1).

Pneumonia is characterized by the recruitment of phagocytic cells, mainly granulocytes, to the site of infection (Skerrett, 1994). The influx of granulocytes into the lung alveolar compartment at the early phase of influenza virus infection was markedly increased in IL-18^-/- mice (Table 1). This vigorous recruitment of granulocytes was associated with proinflammatory stimuli (Greenberger et al., 1996; Tsai et al., 1998), which might be induced by the heavy loading of virus infection in the lung (Fig. 2). The enhanced granulocyte influx into the lung is naturally followed by excessive production of nitric oxide, which is generated mostly by granulocytes and is involved in the pathogenesis of influenza virus-induced pneumonia (Akaike et al., 1996).

No appreciable difference in the local humoral and cellular immune responses upon influenza virus infection could be found between IL-18^-/- mice and wild-type mice (Table 2). Even IL-18^-/- mice cleared progeny virus from the lung completely without delay when the specific immune responses were developed effectively. These results indicate that IL-18 has no effect on the induction and development of influenza virus-specific immunity but controls the innate NK cell and IFN system at an early stage of infection.

This work was supported in part by a grant from the Waksman Foundation of Japan.