Rapid Interferon γ–dependent Clearance of Influenza A Virus and Protection from Consolidating Pneumonitis in Nitric Oxide Synthase 2–deficient Mice

NOS2^−/− mice were generated as described (10). NOS2^−/− and NOS2^+/+ littermates (129/SvEv × C57BL/6 F₂) were bred at the specific pathogen–free unit, John Curtin School of Medical Research. NOS2 genotype was confirmed by PCR (14). Experiments were performed according to institutional guidelines for animal care and use.

Influenza A virus (strain A/ PR/8/34) was directly instilled in 30 μl PBS into the nasal cavity under Avertin (1.5 ml/100 g) anesthesia, and titers were thereafter determined via serial dilutions of lung homogenates plated onto MDCK cell monolayers for detection of PFU (15). The limit of assay detection was 2.0 log₁₀ PFU.

Leukocytes were isolated as outlined (12) from HBSS-perfused and aseptically excised lungs from groups of influenza virus–infected mice (n = 5 per group). Populations consisted of  ∼40–50% T lymphocytes (CD3⁺), 20–30% B lymphocytes (CD45R/B220⁺), and 10–12% macrophages (F4/ 80⁺) as determined by flow cytometry.

Bronchoalveolar lavage fluid (BALF) consisted of washings from intratracheal instillation of 1.0 ml PBS. Nitrite was measured by a diazotization assay (14) with a sensitivity of 4 μM after NO₃⁻ was converted to NO₂⁻ via nitrate reductase.

Influenza A virus–infected and uninfected EL-4 lymphoma cells were used as targets for measurement of anti-influenza A virus–specific, class I MHC–restricted CTL activity of lung parenchymal cells using standard ⁵¹Cr-release assays (16, 17).

A sandwich ELISA (18) was used to measure IFN-γ levels in BALF and from culture supernatants of lung immunocytes (5 × 10⁶; ∼50% CD3⁺) of influenza A/PR/ 8/34 virus–infected mice coincubated for 48 h with influenza virus–infected, mitomycin C–treated (50 μg/ml) genotype-matched lung parenchymal cells (5 × 10⁵). Recombinant MuIFN-γ (Genzyme Corp., Cambridge, MA) served as the standard, with an assay detection limit of 3 U/ml. Con A was added at 4 μg/ml (Sigma Chemical Co., St. Louis, MO).

Formalin-fixed, paraffin-embedded tissues were sectioned (5 μm) and stained with hematoxylin and eosin as outlined (18).

N^ω-methyl-l-arginine (l-NMA; Sigma Chemical Co.) or its inactive d-enantiomer (d-NMA) was administered via daily intraperitoneal injection, and efficacy was verified by BALF nitrite plus nitrate (NO_x) assay. Neutralization of IFN-γ was achieved by injection of 1 mg i.p. anti–MuIFN-γ mAb (clone XMG-6, rat IgG₁ [17, 18]) in 200 μl PBS on days 0, 2, and 4 after infection. Control animals received an isotype-matched mAb against bacterial β-galactosidase (clone GL 113, rat IgG₁).

Unpaired Student's t tests at 95% confidence levels were performed using InStat software, Version 2.00 (GraphPad Software for Science, San Diego, CA).

In previous work, mice treated with anti–IFN-γ antibodies (12, and G. Karupiah, unpublished observations) or lacking IFN-γ (13, and G. Karupiah, unpublished observations) responded to influenza A virus similarly to control mice, even though their capacity for NOS2 expression is known to be impaired (19, 20). Our findings in NOS2^−/− mice not only corroborated the implication that NOS2 is dispensable for the control of influenza A virus in the lower respiratory tract, but also demonstrated that NOS2 is detrimental in this setting. Thus, intranasal inoculation with the virulent A/PR/8/34 strain led to rapid pneumonitis and early mortality in wild-type mice given 300, 400, or 500 hemagglutinin units (HAU) of virus. In contrast, NOS2^−/− mice started to succumb only at 600 HAU (Fig. 1 a).

Influenza A virus titers in the lungs of d-NMA–treated controls given 400 HAU of virus reached 5.6 ± 0.2 log₁₀ PFU by day 6 after infection, whereas in NOS2^−/− mice lung virus titers were below the limit of detection (<2.0 log₁₀ PFU; Fig. 1,b). Surprisingly, l-NMA–treated wild-type mice harbored viral loads similar to their d-NMA– treated littermates (Fig. 1,b) but, unlike the latter, they showed no signs of morbidity and survived through the 21-d period of observation (data not shown). This outcome suggested (a) that NOS2^−/− mice selectively expressed a potent antiinfluenza effector mechanism, and (b) the high output NO pathway may contribute to pathogenesis, since both l-NMA– and d-NMA–treated wild-type mice failed to eradicate the virus but only those given the inactive d-enantiomer proved susceptible, dying between days 7 and 8 after infection, like their untreated counterparts (Fig. 1 a, and data not shown).

Experiments in β₂-microglobulin^−/− mice (21) and Fas^−/− × perforin^−/− chimeras (22) demonstrated the importance of CD8⁺ T lymphocytes for elimination of influenza A virus from the lung. To determine if this effector arm of the immune response contributed to rapid virus clearance in the NOS2^−/− host, we examined influenza-specific CTL activity on day 6 after infection. Lung immunocytes lysed the influenza virus–infected targets equally well irrespective of NOS2 status (Fig. 2,a). However, measurement of local IFN-γ production by these cells (Fig. 2,b) and in BALF (Fig. 2,c) revealed that NOS2^−/− mice exhibited 1.9–3.4-fold higher levels of IFN-γ than wild-type hosts; this increase was independent of antigen load (Fig. 1 b).

Whether such an increase in IFN-γ secretion could account for the reduced viral burden in mutant animals was assessed by intraperitoneal injection with mAb to IFN-γ (18). Assay of virus infectivity levels 6 d after infection demonstrated that IFN-γ–dependent anti-influenza A virus mechanisms were operative in mutant mice as depicted by the high titers comparable to those of l-NMA or d-NMA– treated wild-type controls (Fig. 2,d). In contrast, NOS2^−/− animals receiving an IgG₁ isotype-matched control mAb still retained their capacity to restrict influenza virus replication within the lungs; titers were below the level of detection (Fig. 2 d).

The demise of untreated or d-NMA–treated NOS2^+/+ mice was not due to uncontrolled viral growth, since l-NMA– treated wild-type or anti–IFN-γ–treated NOS2^−/− hosts also displayed overwhelming lung viral burdens yet showed no signs of morbidity throughout the 3-wk experimental period (Figs. 1,b, 2 d, and data not shown). Survival of the latter two groups implied that pulmonary injury in d-NMA– treated animals arose as a consequence of excessive NO production rather than by direct viral cytopathic effects. Indeed, NO_x levels within the BALF of d-NMA–treated NOS2^+/+ mice rose appreciably (∼3.4–4.0-fold) during the course of infection (Fig. 3,a), whereas this increase was absent in mutant mice and inhibited by 77–90% in l-NMA– treated controls (Fig. 3,a). Histopathologic examination revealed massive inflammatory foci and edema within the lungs of all d-NMA–treated wild-type mice as part of the consolidating pneumonitis observed 6 d after infection (Fig. 3 b). Neutrophils and alveolar macrophages were especially prominent. Both are capable of releasing cytotoxic oxidants besides NO (6, 23), such as the superoxide anion (O₂⁻), which can react with the latter to yield more damaging products like peroxynitrite (ONOO⁻ [3]). Akaike et al. have demonstrated the local formation of both ONOO⁻ and NO-hemoglobin in murine influenza infection using nitrotyrosine footprinting and electron spin resonance spectroscopy (6), while NO-thiol adducts also appear elevated in wild-type but not NOS2^−/− mice during tuberculosis (14). Thus, it is conceivable that several NO-derived species acted in concert to cause influenza-associated pathology.

In marked contrast to the severe pneumonitis seen for all d-NMA–treated NOS2^+/+ mice, the majority of mutant animals (∼70%) lacked evidence of pathology. A mild degree of inflammation was observed in the remainder; an example of the latter is depicted in Fig. 3 b. These results are consistent with the idea that NOS2 promotes the development of viral pneumonia. Additional support for this concept comes from recent studies using NOS inhibitors during infection with HSV-1 (7), murine CMV (24), or the influenza A/Kumamoto/Y5/67 (H2N2) mouse adapted strain (6). Inhibition of NOS fully protected mice from HSV-1–dependent lung injury despite a 17-fold increase in viral titers (7). That NO was necessary to control HSV-1 in vivo substantiated earlier experiments in vitro (1, 2). In contrast, in the present work, the functional loss of NOS2 alleviated pneumonitis and greatly improved survival irrespective of whether influenza virus was cleared (NOS2^−/− hosts) or not (NOS2^+/+ mice given l-NMA). Reduced pathology has also recently been noted in NOS2^−/− mice infected with Mycobacterium avium (25) and Toxoplasma gondii (26).

IFN-γ has been implicated as the major cytokine responsible for NOS2 induction during influenza A–induced pneumonitis (6). Aside from IFN-γ's purported proinflammatory action in wild-type mice, we found that heightened IFN-γ release was both associated with and required for influenza A virus clearance in NOS2^−/− animals, an effect unaccompanied by changes in antiviral CTL activity (Fig. 2, a–d). The effect was specific for mutant mice, since l-NMA– treated wild-type mice had comparable IFN-γ levels but failed to limit viral growth. IFN-γ thus appears to cooperate with immune components otherwise subject to suppression by NOS2 and not by NOS1 and/or NOS3. Complete inhibition of NOS2 by disruption of the gene encoding it, rather than the partial inhibition afforded by treatment of NOS2^+/+ mice with l-NMA, also appeared necessary to permit expression of the IFN-γ–dependent antiinfluenza mechanism. Among pertinent candidates for components of this mechanism are MHC class II expression by pulmonary dendritic cells (27) and Th1 cell proliferation plus cytokine production (28, 29). Each could be compromised by NOS2, which can limit T cell responsiveness to IL-2 in the lungs of rats and mice (28) as well as inhibit the release of IL-12 and elevate the expression of TGF-β as shown in Leishmania major–infected NOS2^−/− mice (30). In this manner, the high output NO pathway may regulate IFN-γ secretion via a feedback mechanism, such that once infection is cleared, autotoxic NO production is extinguished.

Beyond this, how NOS2 might influence the underlying molecular events leading to cytokine gene expression and action during antiviral immunity remains speculative. NO can activate or disable protein tyrosine kinases of the Janus kinase (JAK), extracellular signal–regulatory kinase (ERK), and src families, as well as the transcription factor nuclear factor (NF)-κB, and thus may affect cytokine signaling (31, and references therein). Indeed, NOS2^−/− animals have altered levels of NF-κB and signal transducer and activator of transcription (STAT)3 activation in hemorrhage/resuscitation (32) and perturbed IFN-γ release in L. major (28, 29) and influenza A infections (Fig 2, b and c). Cytokine expression could also be regulated by NOS2 at the posttranslational level. For example, S-nitrosylation of the active site cysteine residue in caspases leads to their inactivation (33). If this inhibition were to be relieved in NOS2^−/− mice, abnormally active caspase-1 might more readily promote the release of IL-18 (34, 35), a powerful inducer of IFN-γ.

During infection, NOS2 probably displays all three of its major effects to variable degrees—antimicrobial, inflammatory, and immunosuppressive. The results seen here with influenza A virus–infected mice appear to illustrate one extreme, in which the latter two effects dominate.

We thank Drs. Nicole Baumgarth and Anne Kelso for lung parenchymal cell isolation methods.

This work was supported by the National Centre for HIV Research (G. Karupiah and J.-H. Chen), and by National Institutes of Health grants HL-51967 and AI-34543 (to C.F. Nathan).