Interleukin-22 (IL-22) Production by Pulmonary Natural Killer Cells and the Potential Role of IL-22 during Primary Influenza Virus Infection

Abstract

We set out to test the hypothesis that interleukin-22 (IL-22), a cytokine crucial for epithelial cell homeostasis and recovery from tissue injury, would be protective during influenza virus infection. Recent studies have identified phenotypically and functionally unique intestinal NK cells capable of producing the cytokine IL-22. Unlike gut NK cells that produce IL-22, the surface phenotypes of lung NK cells were similar to those of spleen NK cells and were characteristically mature. With mitogen stimulation, both single and double IL-22- and gamma interferon (IFN-γ)-producing lung NK cells were detected. However, only the IL-22⁺ IFN-γ⁻ lung NK subset was observed after stimulation with IL-23. IL-23 receptor (IL-23R) blocking dramatically inhibited IL-22 production, but not IFN-γ production. Furthermore, we found that NK1.1⁺ or CD27⁻ lung NK cells were the primary sources of IL-22. After influenza virus infection, lung NK cells were quickly activated to produce both IFN-γ and IL-22 and had increased cytotoxic potential. The level of IL-22 in the lung tissue declined shortly after infection, gradually returning to the baseline after virus clearance, although the IL-22 gene expression was maintained. Furthermore, depletion of NK cells with or without influenza virus infection reduced the protein level of IL-22 in the lung. Anti-IL-22 neutralization in vivo did not dramatically affect weight loss and survival after virus clearance. Unexpectedly, anti-IL-22-treated mice had reduced virus titers. Our data suggest that during primary respiratory viral infection, IL-22 seems to a play a marginal role for protection, indicating a differential requirement of this cytokine for bacterial and viral infections.

NK cells are important innate immune effectors that patrol the body for invading pathogens and tumors. Primary biological functions of NK cells include natural cytotoxicity and cytokine generation, through which NK cells directly or indirectly control infections and tumors and regulate the immune system (8). Accumulating evidence has unveiled other novel functions of NK cells that are associated with their anatomic locations. For example, in the uterus, NK cells support reproductive tissue development by providing a variety of cytokines, growth factors, and angiogenic factors (18, 26). The uterine NK cells also demonstrate a unique receptor repertoire, the Ly49 phenotype of which is strikingly different from that of spleen NK cells (39).

Very recently, an NK1.1 low or negative subset of NK cells (CD3⁻ NKp46⁺) has been identified in the intestinal mucosa and found to be capable of making interleukin-22 (IL-22) (7, 24, 31, 32). IL-22 is one of the IL-10 cytokine family members that have been shown to be important in regulating mucosal epithelial cell function, maintaining barrier integrity, and protection from bacterial infections in the gut and lung (4, 43). Interestingly, gut NK cells are distinguished by an immature phenotype, as evidenced by the lack of multiple traditional NK cell markers, such as Ly49A, Ly49D, Ly49C/I, and Ly49G2, and by altered expression of several markers, such as CD122, NK1.1, CD49b (DX5), CD11b, CD27, and CD127, in comparison with spleen NK cells (24, 31, 32). Functionally, gut NK cells lack the capability of gamma interferon (IFN-γ) production and cytotoxicity (24, 31, 32). Taken together, the unique nontraditional features of gut NK cells indicate a distinct developmental process (11, 36) in which they acquire the ability to produce IL-22 and thus are crucial components against intestinal bacterial infections.

In addition to the gut, the respiratory tract is an important mucosal system that can be easily invaded by microorganisms. In the lung, NK cells constitute about 10% of the total resident lymphocytes, a relatively higher percentage than that distributed in most other lymphoid tissues and nonlymphoid tissues (17), indicating potential crucial involvement of NK cells in lung infections. Indeed, lung NK cells are known to be vital for containing numerous pulmonary infections, including those caused by Mycobacterium tuberculosis, Cryptococcus neoformans, Bordetella pertussis, respiratory syncytial virus, and influenza virus (12, 16). The potential mechanism of NK cell defense in lung infections has been attributed to NK cell IFN-γ production and their cytolytic functions. However, IL-22 has been implicated in protection against respiratory infection with Gram-negative bacteria, such as Klebsiella pneumoniae, where IL-22 levels increase after infection (4). Whether lung NK cell production of IL-22 in the context of respiratory virus infection or IL-22 itself is important for viral protection has not been investigated.

In this study, we investigated the phenotypes and IL-22 production potential of lung NK cells in the context of influenza virus infection. The data show that lung NK cells are phenotypically similar to spleen NK cells yet capable of producing IL-22 upon in vitro stimulation and after influenza virus infection in vivo. Unlike gut NK cells, IL-22-producing lung NK cells are capable of making IFN-γ and display cytolytic potential. After influenza virus infection, in spite of the detection of IL-22-producing NK cells in the lung, IL-22 levels actually went down, and mice treated with anti-IL-22 antibodies had reduced virus titers, with little change in disease severity. These observations show that IL-22 serves different roles in bacterial versus virus infections of the lung and suggest that it may be actively regulated to limit proliferation of cells targeted by the influenza virus.

MATERIALS AND METHODS

Animals.

Female C57BL/6 (B6) mice were obtained from The Jackson Laboratory and used between the ages of 8 and 10 weeks. All animals were housed in the University of Rochester Vivarium facilities under specific-pathogen-free conditions, using microisolator technology. Intranasal inoculation with influenza virus A/HK×31 (H3N2) and influenza virus A/PR/8/34 (PR8, H1N1) was as described previously (30). Virus titers of the collected samples were analyzed with MDCK cell-based plaque assay (9). All animal experiments were performed according to the protocol approved by the university animal committee.

Antibodies and reagents.

APC-Cy7-conjugated anti-CD3, fluorescein isothiocyanate (FITC)-conjugated anti-CD11b, FITC-conjugated anti-Ly49C and anti-Ly49I, FITC-conjugated anti-Ly49D, FITC-conjugated anti-Ly49A (B6), PerCPCy5.5-conjugated anti-CD69, and phycoerythrin (PE)-conjugated anti-CD127 were purchased from BD Pharmingen (San Jose, California). Alexa 647-conjugated anti-NCR1, PE- or FITC-conjugated anti-CD107a, FITC-conjugated anti-Ly49G2, FITC-conjugated anti-CD122, FITC-conjugated anti-CD49b (DX5), PerCPCy5.5-conjugated anti-NK1.1, FITC- or PerCPCy5.5-conjugated anti-IFN-γ, and unconjugated anti-CD16/32 were from eBioscience (San Diego, CA). PerCPCy5.5-conjugated anti-CD27 was from Biolegend (San Diego, CA). PE-conjugated anti-IL-22 antibodies and unconjugated anti-IL-23R were obtained from R&D Systems (Minneapolis, MN). Live/Dead fixable violet fluorescent reactive dye was from Molecular Probes (Invitrogen, Eugene, OR). Phorbol myristate acetate (PMA), ionomycin, monensin, and Histopaque 1083 were ordered from Sigma (St. Louis, MO). Fixation and permeabilization were performed using a BD Cytofix/Cytoperm kit from BD Bioscience (San Jose, CA).

Cell preparation and stimulation.

Lungs and spleens were harvested and minced. Spleen samples were depleted of red blood cells (RBC) by using a buffered ammonium chloride solution for 5 min. Lung single-cell suspensions were obtained by pressing the organs through a 200-gauge wire mesh and filtered through a 90-μm nylon mesh. Lung lymphocytes were isolated at the interface of a 30-min centrifugation step at 1,500 × g with Histopaque 1083. Cells were counted with trypan blue exclusion. Cell samples either blocked or unblocked with 10 μg/ml anti-IL-23R (10⁵ per well) were stimulated with PMA and ionomycin (PMA-ionomycin) in a final concentration of 100 ng/ml for PMA and 500 ng/ml for ionomycin for 5 h at 37°C, with monensin (5 μg/ml) added in the last 3 h.

Antibody staining.

Freshly isolated or cultured cells were washed with staining buffer (phosphate-buffered saline [PBS]-1% fetal bovine serum [FBS]) and blocked with unlabeled anti-CD16/32 for 20 min, followed by staining with Live/Dead violet dye and respective antibodies for 30 min at 4°C. For cytokine staining, cells were then fixed with 100 μl of Cytofix/Cytoperm for 20 min at 4°C, followed with two washes using permeabilization-wash buffer (perm/wash buffer; BD Biosciences, PaloAlto, CA). Intracellular staining for IFN-γ and IL-22 was performed for 30 min at 4°C. Cells were then washed twice with perm/wash buffer and resuspended in staining buffer before samples were run in the LSRII machine (BD Biosciences, San Jose, CA). All fluorescence-activated cell sorter (FACS) data were analyzed using FlowJo software (Tree Star, San Carlos, CA).

IL-22 real-time reverse transcription-PCR (RT-PCR).

Lung tissue RNAs were extracted using the RNeasy mini kit (Qiagen), and cDNAs were transcribed using the Omniscript RT kit (Qiagen). Real-time PCR was performed using TaqMan universal PCR master mix on the Applied Biosystems Prism 7700 sequence detection system. Primers and probes to IL-22 (Mm00444241_m1) and hypoxanthine guanine phosphoribosyl transferase, HRPT (Mm00446968_m1), were purchased from Applied Biosystems. The average threshold cycle (C_t) value of each sample was obtained by SDS RQ Manager software (Applied Biosystems). Relative expression levels of IL-22 were obtained after normalization with HRPT by using ΔΔC_t methods. The fold change of IL-22 after influenza virus infection was calculated by comparison with its level in uninfected lungs.

NK cell depletion.

One hundred micrograms of anti-NK1.1 antibody (clone PK136; eBioscience) was administered intraperitoneally to each mouse 12 h before intranasal influenza virus infection. The efficacy of depletion was monitored by FACS analysis with CD3 and NKp46 staining. Absence of NK cells in the lung was observed for up to 10 days (data not shown).

IL-22 and IL-23 detection and in vivo neutralization.

IL-22 or IL-23 in the homogenized lung supernatants and bronchial lavage samples was quantified by mouse IL-22 or IL-23 enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems) according to the product instructions. For anti-IL-22 treatment, we doubled the injection dose from 50 μg (4) to 100 μg of neutralizing goat anti-mouse IL-22 (R&D Systems) to ensure satisfied neutralization. Animals were administered the anti-IL-22 treatment intraperitoneally immediately after intranasal influenza virus infection (day 0) and 5 days after infection (day 5).

Statistical analysis.

Statistical significance was evaluated using the two-tailed, unpaired Student t test when comparing the appropriate groups. A P value of <0.05 was considered statistically significant.

RESULTS

Lung NK cells are phenotypically mature.

Although the lung and the gut are both mucosal surfaces, they have many unique features. Similarities and differences among the NK cell populations at these two sites are not well defined. Compared to spleen CD3⁻ NKp46⁺ NK cells, gut NK cells have no or very low expression of a number of classical NK cell markers, which include CD122, CD49b (DX5), CD11b, Ly49A, Ly49D, Ly49C/I, and Ly49G2 (24, 31, 32). They also show decreased expression of NK1.1 and CD27 and increased expression of CD127 and CD69 in comparison with those shown by spleen NK cells (24, 31, 32). CD3⁻ NKp46⁺ NK cells among live lymphocytes isolated from lung or spleen were examined for the expression of individual markers, including CD122, CD49b (DX5), CD11b, Ly49A, Ly49D, Ly49C/I, Ly49G2, NK1.1, CD27, CD127, and CD69. The percentages of the lung NK cell Ly49 receptor repertoire, including Ly49A, -D, -C/I, and -G2, are essentially the same as those of spleen NK cells (Fig. 1A), although lung NK cells have higher mean fluorescence intensities (MFI) for Ly49D and Ly49C/I than do spleen NK cells. Expression of CD122, CD49b, and CD11b on lung NK cells was not only preserved but appeared to be even more homogeneous than that of those on spleen NK cells. NK1.1 expression on lung NK cells also appeared to be more uniform. As reported by Hayakawa and Smyth (20), CD27 expression was downregulated in the lung NK cells compared to that in spleen NK cells (Fig. 1B). Expression of CD127 and CD69, which are upregulated on the gut NK cells, was decreased or unaltered on the lung NK cells compared to that on spleen NK cells, respectively (Fig. 1C). These results suggest that lung NK cells bear more mature phenotypes.

FIG. 1.

Lung NK cells are phenotypically mature. Freshly isolated lung and spleen cells were stained and gated for CD3⁻ NKp46⁺ NK cells and analyzed for the expression of a panel of markers that are absent (A), downregulated (B), or upregulated (C) on gut NK cells compared to on spleen NK cells (24, 31). In each overlay, gray-filled and unfilled histograms represent the expression of individual markers on spleen and lung NK cells, respectively. The numbers on each plot represent the frequency of each molecule from gated NK cells. Data are representative of the results for three independent experiments.

Lung NK cells produce IL-22 upon in vitro stimulation and require IL-23R signaling.

IL-22 is thought to be important for maintaining the integrity of the epithelium during infection. To examine the potential ability of lung NK cells to produce IL-22, Ficoll-isolated mouse lung single-cell suspensions were stimulated with PMA-ionomycin and processed for intracellular staining of IL-22 and IFN-γ and FACS analysis. Dead cells were excluded by Live/Dead violet dye, and NK cells were defined as CD3⁻ NKp46⁺ cells. As shown in Fig. 2A, upon stimulation, lung NK cells produced IFN-γ, and about 10% of lung NK cells were induced to express IL-22. Interestingly, about 25% of the total IL-22-producing lung NK cells were also positive for IFN-γ. As IL-23 can directly promote polarized Th17 cells and gut NK cells to produce IL-22 (7, 24, 31, 42), we also stimulated lung cells with IL-23. Under these conditions, IL-22 was clearly produced by lung NK cells (Fig. 2A and B); however, similar to gut NK cells, no IFN-γ could be detected in lung NK cells when using IL-23 alone (Fig. 2A). To further investigate whether production of IL-22 by lung NK cells is initiated through IL-23 receptor (IL-23R)-mediated signaling, anti-IL-23R blocking antibody or isotype control antibody was added into lung cell cultures before stimulation. IL-22 production by lung NK cells was almost completely inhibited by anti-IL-23R, either with PMA-ionomycin or IL-23 (Fig. 2B) stimulation. However, IFN-γ production by lung NK cells after PMA-ionomycin stimulation was not affected with anti-IL-23R blocking (Fig. 2B). Taken together, these results suggest that lung NK cells have the ability to produce IL-22 through IL-23R-mediated signaling, but unlike gut NK cells, IL-22-producing lung NK cells also have the potential to make IFN-γ when the appropriate signals are presented.

FIG. 2.

Production of IL-22 by lung NK cells in vitro. Lung cells were isolated and stimulated with either PMA-ionomycin (PMA/Iono) or IL-23 in the absence (A) or presence (B) of anti-IL-23R blocking antibody as described in the text. Gated live CD3⁻ NKp46⁺ lung NK cells were analyzed for intracellular IFN-γ and IL-22. Data are representative of the results for at least three independent experiments.

NK1.1⁺ and CD27⁻ lung NK cells are major IL-22 contributors.

Although gut NK cells could produce IL-22, this ability is limited to NK1.1-negative or low subsets (24, 31). Since CD3⁻ NKp46⁺ lung NK cells are nearly all NK1.1⁺ (Fig. 1B and 3A), it is not surprising that NK1.1+ lung NK cells were the major source of IFN-γ or IL-22 (Fig. 3A), although we did find that the limited number of NK1.1⁻ lung NK cells also clearly produced IL-22 after stimulation either with PMA-ionomycin or IL-23 (Fig. 3A).

FIG. 3.

IL-22 production by lung NK cell subsets. Lung cells were isolated and stimulated with either PMA-ionomycin or IL-23. Live CD3⁻ NKp46⁺ lung NK cells were gated for analysis of NK1.1 and IFN-γ or IL-22 (A). Alternatively, lung NK cells were analyzed for CD27 expression after PMA-ionomycin stimulation in comparison with isotype control (shaded histogram) (B), and CD27^+/− subsets were further examined for IFN-γ and IL-22 expression (C). Data are representative of the results for two independent experiments.

The CD27⁺ NK cell subset has been proposed to have high IFN-γ production potential (19); however, it is not known whether this is true for other cytokines. CD27⁺ and CD27⁻ lung NK cells were compared for both IFN-γ and IL-22 (Fig. 3). In agreement with previous findings on spleen NK cells (20), both CD27⁺ and CD27⁻ lung NK cells were able to produce IFN-γ after PMA-ionomycin stimulation; however, a higher proportion of CD27⁺ lung NK cells produced IFN-γ (Fig. 3C). In contrast, their abilities to generate IL-22 were comparable (Fig. 3C), irrespective of coproduction of IFN-γ (IL-22⁺ IFN-γ⁻, 9.8% for the CD27⁺ subset and 10.1% for the CD27⁻ subset; or IL-22⁺ IFN-γ⁺, 5.1% for the CD27⁺ subset and 2.1% the CD27⁻ subset). Since the CD27⁺ subset was composed of only about 30% of the total lung NK cells (Fig. 3B), IL-22 production by lung NK cells was primarily derived from CD27⁻ lung NK cells.

IL-22 generation by lung NK cells after influenza virus infection.

After the finding that lung NK cells have the ability to generate IL-22 upon in vitro stimulation, we decided to test further whether lung NK cells could produce IL-22 in vivo after intranasal infection with PR8 (H1N1) influenza virus. Lung single-cell suspensions were collected at different time points after nonlethal PR8 infection and immediately subjected to surface and intracellular staining without any in vitro restimulation. As seen in Fig. 4A, the level of IFN-γ expression by NK cells peaked at day 2 after infection and declined thereafter, consistent with the prevailing idea that NK cells are among the early effectors that are activated rapidly after infections. At day 2 of infection, neither the total number of lung lymphocytes nor the number of lung NK cells changed noticeably (Fig. 4B and C), suggesting that rapidly activated cytokine-producing NK cells at day 2 are likely lung resident NK cells. At day 2, not only IFN-γ but also enhanced levels of IL-22 were clearly detected on lung NK cells (Fig. 4A). More interestingly, in agreement with our in vitro stimulation results, a small number (about 20%) of IL-22-producing lung NK cells were also positive for IFN-γ at day 2 after infection (Fig. 4A), and a few IL-22⁺ IFN-γ⁺ lung NK cells remained detectable at day 4. Similar results were seen when mice were infected with a lethal dose of PR8 virus or X31 virus that causes only mild disease in mice (data not shown). Without ex vivo stimulation, neither IFN-γ nor IL-22 could be detected in the spleen NK cells concurrently examined with lung NK cells after infection (data not shown). These data indicate in vivo activation of lung NK cell cytokine production in response to influenza viral infection.

FIG. 4.

Production of IL-22 by lung NK cells in vivo. Mice were infected with 5 PFU influenza PR8 virus, and lungs were collected at different time points. Lung single-cell suspensions were purified and counted before surface and intracellular IFN-γ and IL-22 staining was performed and analyzed (A). The absolute number of lung NK cells was obtained by multiplying the total number of lung lymphocytes counted with the percentage of NK cells (B). At each time point, three to five mice were used. Data are representative of the results for two independent experiments.

Influenza virus-induced IL-22-producing lung NK cells have cytotoxic potential.

It is known that shortly after PR8 influenza virus infection, lung NK cells acquire increased killing activity against YAC-1 tumor cells (34). In order to examine whether IL-22-producing lung NK cells have cytotoxic potential during PR8 infection, we measured lung NK cell cytotoxic potential by analyzing the surface level of CD107a, which is correlated with NK cell cytotoxicity and is widely used to indicate cytotoxic potential (1-3, 6, 15, 37). At day 2 of PR8 virus infection, lung NK cells showed an increased level of CD107a, but no obvious increase in CD107a was seen on spleen NK cells analyzed concurrently (Fig. 5A and B). Increased expression of CD107a on lung NK cells peaked around day 6 (Fig. 5A) and rapidly declined after day 8 (data not shown). To see whether IL-22-producing lung NK cells have cytotoxic potential, day-2 lung NK cells were analyzed for both CD107a and IL-22. As shown in Fig. 5C, day-2 lung NK cells had not only increased CD107a expression but also increased IL-22 expression, in comparison with day-0 lung NK cells. A portion of IL-22-producing lung NK cells were also positive for CD107a, suggesting that IL-22-producing lung NK cells could be cytotoxic in vivo after influenza virus infection.

FIG. 5.

Cytotoxic potential of IL-22-producing lung NK cells. Mice were infected with 5 PFU influenza PR8 virus determined by staining and tissues were collected at different time points for analysis. NK cell cytotoxicity was measured with the expression of CD107a in both lung (A) and spleen (B) samples immediately after cell preparation by FACS. IL-22 production by lung NK cells was also determined by staining and analyzed together with CD107a (C). Data are representative of the results for two independent experiments.

Total IL-22 production and relative contribution by NK cells after influenza virus infection.

Although we observed that lung NK cells could be activated to produce IL-22 during influenza virus infection (X31 or PR8), the IL-22 protein level and the relative contribution of NK cells to IL-22 production are not known. By using ELISA, we showed that the total IL-22 protein level in the lung decreased after day 2 of the acute infection and gradually increased to approach the baseline after 8 days (Fig. 6A), when virus has typically been cleared (14). IL-22 remained undetectable in the airway lavage samples during the whole course of primary infection (data not shown). One possibility for the observed reduction of IL-22 was that the overall expression of the IL-22 gene in the lung was inhibited after infection. To test this, IL-22 real-time RT-PCR was performed. As demonstrated in Fig. 6B, after influenza virus infection, IL-22 gene transcript was not inhibited compared to its expression at day 0. In fact, at the four time points examined, various levels of increased expression of total IL-22 gene were observed, with the highest induction seen at day 2 of infection, followed by some decrease at day 4, but increases to different levels at day 8 and day 12 after infection.

FIG. 6.

Lung IL-22 expression after influenza virus infection. Mice were infected with 5 PFU PR8 influenza virus or 3 × 10⁵ 50% egg infective doses (EID₅₀) of X31 virus as described in Material and Methods. (A) Kinetics of IL-22 level in the lung after influenza virus infection. At each time point as indicated in the figure, three mice were used. (B) Lung IL-22 gene expression after influenza virus infection in comparison to noninfection control. Lung total RNA was extracted and RT-PCR performed as described, and the relative amount of IL-22 was obtained after normalization with the HRPT housekeeping gene. Data are averages for three mice at each time point analyzed. (C) IL-22 levels in the lungs of uninfected mice after anti-NK1.1 antibody depletion. Mice were injected with anti-NK1.1 or isotype control antibodies as described in the text, and lungs were collected at different times for IL-22 detection. Data presented are average values for three mice for each group. (D) IL-22 levels in the lungs of PR8-infected mice after anti-NK1.1 antibody depletion. Mice were depleted with anti-NK1.1 antibodies 12 h before infection, and lung samples were collected for IL-22 measurement at the indicated time. Data presented are average values for six mice for each group.

Our in vitro study has demonstrated that IL-23 could stimulate lung NK cell to generate IL-22. To further test whether the induction of IL-22 gene expression and the production of IL-22 protein were correlated with IL-23, we examined IL-23 protein level in lung by using ELISA methods as described above. Surprisingly, IL-23 was not detected from the tissue collected at different time points after infection (data not shown). This could be due to the fact that the IL-23 level is too low to be detected or that IL-23 production is not initiated in the lung. Alternatively, it is possible that IL-23 is not the only cytokine that can signal IL-22 production. The latter possibility is also suggested by recent studies (7, 13, 41) that have reported that other cytokines, like IL-15, IL-12, and IL-18, could also induce NK cells to produce IL-22.

Since lung NK cells could produce IL-22 early in influenza virus infection (Fig. 4A), we intended to determine the relative contribution of lung NK cells to the total lung IL-22 level. Lung NK cells and those in other tissues were depleted with a single injection of 100 μg of anti-NK1.1 antibody, a regimen that caused them to remain depleted for at least 10 days (data not shown). Depletion in uninfected mice reduced but did not eliminate detectable IL-22 in the lung over a period of several days (Fig. 6C). To further examine the effect of NK cell depletion on IL-22 production after influenza virus infection, we injected anti-NK1.1 antibody 12 h before infection. At day 2 after infection, a time point at which we see peak intracellular levels of IL-22 in lung NK cells, IL-22 protein levels in the lung were significantly reduced (Fig. 6D). However, by 10 days of infection, total lung IL-22 levels were similar in depleted and intact mice (Fig. 6D), suggesting that NK cells contribute to lung IL-22 production mostly in the early phase of infection. Collectively, the data demonstrate that lung NK cells contribute to lung IL-22 production but are not the only source.

Neutralization of IL-22 did not impact influenza virus infection severity.

Given the protective effect of IL-22 in bacterial infections of the lung and gut and the potential for IL-22 production during influenza virus infection, we investigated the importance of IL-22 for protection from influenza virus infection. Though our data showed a reduction in detectable IL-22 after infection, gene expression was maintained to a certain degree. This raised the possibility that active consumption of the cytokine by damaged epithelium was also contributing to the observed reduction in protein levels. If so, we reasoned that early inhibition of IL-22 would increase susceptibility to severe disease even after nonlethal virus infection. To test this prediction, we injected 100 μg of anti-IL-22 neutralizing antibody (4, 43) twice at days 0 and 5. Contrary to our expectations, there was no difference in mortality (data not shown) and only a slight, though not significant, increase in weight loss after either low-dose PR8 (Fig. 7B) or X31 infection (data not shown) compared to that in isotype treatment controls. Unexpectedly, treatment with anti-IL-22 reduced the titer of virus, measured in lung or trachea (Fig. 7A) at day 5 after infection. These observations indicate that IL-22 is not critical to protection from acute influenza virus infection and may in fact sustain the respiratory epithelial cells and promote influenza virus replication in these target cells.

FIG. 7.

Neutralization of IL-22 after influenza virus infection. Mice were infected with 5 PFU PR8 influenza virus and treated with either anti-IL-22 or isotype control antibodies as described in Materials and Methods. (A) Virus titers of lung and trachea samples collected from anti-IL-22- or isotype control-treated animals at day 5 of infection. Data are averages of the results for three mice in each group analyzed. (B) Change in body weight of infected animals with anti-IL-22 or isotype control treatment was calculated as the percentage of initial weight. Data presented are average values for five mice for each group and representative of the results for two independent experiments.

DISCUSSION

We set out to investigate the hypothesis that the lung contained a population of NK cells analogous to the IL-22-producing NK cells of the gut and that IL-22 would be an important mediator of protection in a model of respiratory virus infection. These analogies were drawn from observations made in models of lung and gut infection with bacteria (4, 43), with the link from the ability of IL-22 production by NK cells in the lung to the role of IL-22 in respiratory viral infection not yet demonstrated. In contrast to other anatomical locations, the NK cells in the lung were not unusual in phenotype, but they have the ability to generate IL-22 with distinct features. Unlike the bacterial models, we did not find further evidence that IL-22 was critical to controlling influenza virus infection. If anything, inhibition of IL-22 reduced viral burden, suggesting that antagonism of its production could be a normal consequence of the virus infection. Reduced levels of detectable IL-22 in lung homogenates and undetectable amounts in the lavage fluid during the infection support this idea.

In the mammalian body, there are three major mucosal systems: the oral-gastrointestinal, the respiratory, and the genitourinary systems. Each of these mucosal surfaces is susceptible to infection by unique routes. Localized mucosal innate immune responses play a key role for the initial suppression of pathogen replication, propagation, and spread. These responses could directly initiate antigen recognition and presentation for the induction of adaptive responses through which infections are usually controlled. In the respiratory mucosal systems, NK cells together with macrophages and dendritic cells (DCs) are major components of innate surveillance and defense effectors. Although it has been noted for a while that NK cells in the lung are important mediators for controlling some tumors and infections (28, 35), detailed information regarding lung NK cell phenotypes is scarce. Lung NK cells express NK1.1, a marker traditionally used for defining NK cells in certain mouse strains (22), and are used in inhibition or depletion type studies to reveal antitumor or antipathogen lung NK cell cytotoxicity (21, 23, 34). Recently, a novel NK cell marker, NKp46 or NCR1, has been found to be expressed in all strains of mouse NK cells, and CD3⁻ NKp46⁺ NK cells from various tissues, including lung, are uniformly NKp46⁺ (37). NKp46 is known to be able to recognize the influenza virus hemagglutinin expressed on infected cells (27), and mice deficient in this receptor do not control influenza virus infection well (16). Our phenotypic results in this study showed that lung NK cells defined by CD3⁻ NKp46⁺ cells are essentially uniformly NK1.1⁺. In line with previous reports (23), high levels of DX5 (alpha-2 integrin) were detected on lung CD3⁻ NKp46⁺ NK cells, but compared to spleen CD3⁻ NKp46⁺ NK cells, lung CD3⁻ NKp46⁺ NK cells displayed increased homogeneity of DX5. Similarly, as previously reported (20), lung CD3⁻ NKp46⁺ NK cells had decreased expression of CD27 compared to spleen NK cells. We also found that several other traditional NK cell markers, like CD122, CD11b, and Ly49 inhibitory or activation receptors, were expressed on lung NK cells. These results indicate that NK cells in the respiratory mucosal system are developmentally mature (10, 40).

Functionally, lung NK cells have been well known for cytotoxicity and IFN-γ production, but other functions have not been closely examined. Multiple recent studies report that gut mucosal NK cells that bear immature NK cell phenotypes could generate the cytokine IL-22 for protection from bacterial infections (7, 24, 31, 32, 41). IL-22 is a member of the IL-10 family of cytokines that also includes IL-10, IL-19, IL-20, IL-24, IL-26, IL-28, and IL-29, all of which bear a homologous α-helical secondary structure (38). The receptor for IL-22 (IL-22R) is a heterodimer complex that is composed of IL-22R1 and IL-10R2 (38). While IL-10R2 is widely expressed on major immune cells, expression of IL-22R1 is restricted on the epithelium of various tissues, like digestive tissue, skin, kidney, and lung. Since the specificity of IL-22-induced signals is conferred by IL-22R1 (5), the biological effect of IL-22 appears to target tissue epithelium. Indeed, accumulating data show that IL-22 could stimulate the expression of genes important for epithelium migration, repair, and wound-healing responses, supporting the crucial role of IL-22 in epithelial-barrier function (5, 29). Until now, whether a population of NK cells analogous to the gut IL-22 NK cells existed in the lung mucosa was not clear.

In this study, we evaluated the IL-22 production potential of NK cells in the respiratory mucosal system. We found that lung NK cells could produce IL-22 in vitro and in vivo. However, we observed that IL-22-producing lung NK cells could also generate IFN-γ after PMA-ionomycin stimulation. Thus, although mucosal NK cells in both the respiratory tract and the intestine could generate IL-22, only lung NK cells producing IL-22 also have the ability to make IFN-γ (24, 31, 32). Interestingly, the presence of IL-22⁺ IFN-γ⁻, IL-22⁺ IFN-γ⁺, and IL-22⁻ IFN-γ⁺ lung NK subsets reflects the IL-22 and IFN-γ production pattern of CD4 Th17 cells under various stimulations (25, 33). Similar to what occurred with intestinal NK cells, IL-22 production by lung NK cells was detected after stimulation with IL-23. Blocking with anti-IL-23R essentially inhibited the IL-22 production by either PMA-ionomycin or IL-23 stimulation, without an effect on IFN-γ production. These results suggest that IL-22 and IFN-γ production by lung NK cells possibly use different pathways. The data also suggest that IL-23 is required for IL-22 production by NK cells, regardless of the stimulation.

Contrary to our expectations, IL-22 levels decreased significantly during the course of influenza virus infection, a result out of line with events during respiratory infection with Gram-negative bacteria (4). Similarly, the effects of blocking IL-22 function were negligible during primary influenza virus infection. These results negate the hypothesis that IL-22 is critical for controlling influenza virus. Although an IL-22-producing subset of NK cells exists in the lung during influenza virus infection, with detectable intracellular protein and evidence of partial contribution to the overall lung IL-22 level, as suggested by the NK cell depletion study, it seems unlikely that this is an important contributor to the protective effect exerted by lung NK cells against the virus. Why these results differ is not clear, though there are many possibilities. Systemic spread of influenza virus does not occur in these experimental models of infection, and unlike bacterial infection, influenza virus infection does not cause a lethal systemic inflammatory response (sepsis). The main difference in outcomes may thus be determined by the ability of the pathogen to spread outside the lung. It would be interesting to repeat these studies with an influenza virus capable of systemic replication, like the highly pathogenic H5 viruses. Many other questions remain, including whether we would observe a more significant phenotype in IL-22-deficient mice. Additional studies are necessary to investigate this intriguing possibility. Furthermore, it may be important to determine whether the downmodulation of IL-22 in the lung during the virus infection is an active process intended to reduce the availability of target cells, though at this time, we can only speculate on how this mechanism might operate.