Delayed antiviral plus immunomodulator treatment still reduces mortality in mice infected by high inoculum of influenza A/H5N1 virus

Abstract

The mortality of human infection by influenza A/H5N1 virus can exceed 80%. The high mortality and its poor response to the neuraminidase inhibitor oseltamivir have been attributed to uncontrolled virus-induced cytokine storm. We challenged BALB/c mice with 1,000 LD₅₀ of influenza A/Vietnam/1194/04. Survival, body weight, histopathology, inflammatory markers, viral loads, T lymphocyte counts, and neutralizing antibody response were documented in infected mice treated individually or in combination with zanamvir, celecoxib, gemfibrozil, and mesalazine. To imitate the real-life scenario, treatment was initiated at 48 h after viral challenge. There were significant improvements in survival rate (P = 0.02), survival time (P < 0.02), and inflammatory markers (P < 0.01) in the group treated with a triple combination of zanamivir, celecoxib, and mesalazine when compared with zanamivir alone. Zanamivir with or without immunomodulators reduced viral load to a similar extent. Insignificant prolongation of survival was observed when individual agents were used alone. Significantly higher levels of CD4⁺ and CD8⁺ T lymphocytes and less pulmonary inflammation were also found in the group receiving triple therapy. Zanamivir alone reduced viral load but not inflammation and mortality. The survival benefits of adding celecoxib and mesalazine to zanamivir could be caused by their synergistic effects in reducing cytokine dysfunction and preventing apoptosis. Combinations of a neuraminidase inhibitor with these immunomodulators should be considered in randomized controlled treatment trials of patients suffering from H5N1 infection.

The mortality of patients suffering from pneumonia and multiorgan involvement caused by influenza A/H5N1 virus (H5N1) varies between 45% and 81% since earlier reports (1, 2). Subsequent use of oseltamivir has not reduced the mortality associated with this virus. The unsatisfactory outcome of its treatment was attributed to either deficiencies in antiviral administration or the induction of a severe cytokine storm (3). The poor response to oseltamivir can also be the result of delayed initiation of treatment because of the nonspecific initial manifestations of H5N1 infection, high initial viral load, poor oral bioavailability of oseltamivir in the seriously ill, lack of i.v. preparations of oseltamivir, and the emergence of resistance during therapy (4, 5). Attempts to use antiinflammatory doses of corticosteroids to control excessive inflammation were associated with severe side effects without any improvement in survival (6). Moreover, cytokine and chemokine knockout mice or steroid-treated wild-type mice did not have survival advantage over wild-type mice after viral challenge (7). This paradox can be explained if both a high initial viral load and the commensurate degree of excessive inflammation are important in the pathogenesis and outcome of H5N1 infection. Here, we tested the hypothesis that the combination of a parenterally administered neuraminidase inhibitor, zanamivir, together with the cyclooxygenase-2 (COX-2) inhibitor celecoxib and mesalazine could be effective in reducing mortality. To imitate the clinical situation, we delayed combination therapy for 48 h after challenging the BALB/c mice with an inoculum of 1,000 LD₅₀ of a highly virulent influenza A virus, human isolate A/Vietnam/1194/04. Our results demonstrate that combination therapy consisting of an inhibitor of the viral neuraminidase (zanamvir) and two inhibitors of inflammation (celecoxib and mesalazine) greatly increased the survival rate of mice infected with a highly pathogenic strain of influenza A/H5N1 virus.

Results

All mice survived with early institution of i.p. zanamivir treatment (Fig. 1A). The survival rate of mice was decreased to 13.3% (2/15) if the treatment with zanamivir was delayed for 48 h, although the mean survival time was prolonged to 10.7 ± 1.6 days compared with 6.6 ± 1.6 days in controls (Fig. 1B). As expected, all PBS-treated controls died, whereas all mice on immunomodulators alone died with a trend toward increased mean survival time to ≈8.5 days for mice given celecoxib or mesalazine and ≈9.5 days for those given both celecoxib and mesalazine, but only ≈7.5 days for those given gemfibrozil alone or both celecoxib and gemfibrozil. Therefore, we did not select gemfibrozil for further study. Single use of any of these immunomodulators did not confer survival benefit. However, when zanamivir was combined with both immunomodulators, the survival rate increased to 53.3% (8/15) (P = 0.02) and the mean survival time increased to 13.3 days (P = 0.0179) compared with zanamivir alone (survival rate 13.3% and survival time 8.4 days). The body weight of all infected mice steadily decreased to a minimum at day 11 and then increased again for those that survived (Fig. 1C).

Fig. 1.

Survival times, survival rates, and body weight for infected mice treated with zanamivir, celecoxib, mesalazine, and gemfibrozil. (A) Shown is the survival rate and time of the mice (five mice per group) treated with zanamivir (Z), celecoxib (C), mesalazine (M), gemfibrozil (G), celecoxib/mesalazine (C+M), celecoxib/gemfibrozil (C+G), and PBS (control) at 4 h postchallenge. (B) Survival time and rate of the mice (10–15 mice per group) treated with zanamivir (Z), zanamivir/celecoxib (Z+C), zanamivir/mesalazine (Z+M), zanamivir/celecoxib/mesalazine (Z+C+M), and PBS at 48 h postchallenge were monitored for 21 days. (C) Body weights of the mice were monitored for 21 days (survival mice) or until death.

Significant decrease (>2.5 logs) of viral titers in tracheal-pulmonary lavage (TPL) by TCID₅₀ or copies of viral RNA genomes in lung tissues by real-time quantitative RT-PCR was found in groups treated by zanamivir with or without immunomodulators at days 6 and 8 postchallenge (Fig. 2). Levels of inflammatory markers IL-6, IFN-γ, TNF-α, macrophage inflammatory protein 1 (MIP-1), and leukotriene assayed by enzyme immunoassays were significantly higher in TPL obtained from the mice treated with zanamivir alone and controls than those treated by triple therapy (P < 0.01 or 0.05) or uninfected normal mice (Fig. 3). However, IL-1 levels were only slightly lower in those treated with zanamivir alone and controls (P > 0.05), whereas prostaglandin E₂ (PGE₂) levels were found to be significantly higher in the samples collected at day 8 postchallenge from the group receiving triple therapy (Fig. 3). As expected, their serum cytokine and chemokine changes were similar to those of their TPL [supporting information (SI) Fig. S1]. Furthermore, levels of both CD4⁺ and CD8⁺ T lymphocytes were significantly higher in the blood taken at days 6 and/or 8 from the mice given triple therapy than those taken from zanamivir-treated and PBS control mice (Fig. 4A).

Fig. 2.

Detection of viral load in infected mice treated with zanamivir, celecoxib, and mesalazine. (A) Titers of released virus in TPL collected at the indicated days from mice treated with zanamivir alone (Z), zanamivir/celecoxib/mesalazine (Z+C+M), and PBS, which were started at 48 h postchallenge, as measured by TCID₅₀. The detection limit (undetectable) is 1:20. (B) Viral RNA copies in lung tissue from the above mice were determined by real-time RT-PCR and normalized by β-actin. The P values between groups Z+C+M and Z or PBS are indicated.

Fig. 3.

Detection of proinflammatory cytokines, chemokines, prostaglandins, and albumin in TPL. Concentrations of IL-1, IL-6, IFN-γ, IFN-α, MIP-1, PGE₂, leukotrienes, and albumin in TPL collected from mice treated with zanamivir alone (Z), zanamivir/celecoxib/mesalazine (Z+C+M), untreated control (PBS), and uninfected (normal) mice at the indicated days were determined by ELISA and compared between different groups. Their P values are shown.

Fig. 4.

T lymphocyte counts in peripheral blood and histopathologic changes in lungs. (A) Numbers of CD3⁺/CD4⁺ and CD3⁺/CD4⁺ T lymphocytes in 10,000 blood cells taken from the mice in the indicated days were counted by flow cytometry, and the P values between zanamivir alone (Z), zanamivir/celecoxib/mesalazine (Z+C+M), and PBS groups are shown. (B) Histopathologic changes in mouse lung tissues collected at the indicated days postinfection are shown. Representative histologic sections of the lung tissues from these mice and uninfected mice (Normal) were stained with H&E (original magnification: ×100). Inflammatory infiltrate and alveolar damage are seen as thickening of the alveolar septum with obliteration of some alveolar spaces at this magnification. (C) Viral infection in lung of the mice was further demonstrated by immunohistochemical staining. Positivity is indicated by brown staining in the cytoplasm. Representative histologic sections of the lung tissues taken from these mice and uninfected mice (Normal) at the indicated day were stained with an antiinfluenza NP mAb (Original magnification: ×400.)

The degree of lung damage as evidenced by increased albumin levels in the TPL (Fig. 3) and higher elastase activity in the TPL (Fig. S1) was significantly lower in groups treated by the combination therapy compared with the groups treated by zanamivir alone (P < 0.01) or PBS (P < 0.03). Histopathologic examination further showed that the alveolar damage and interstitial inflammatory infiltration in mice treated by the combination were much less severe than those treated by zanamivir alone (Fig. 4B). Immunohistochemical staining with antiinfluenza nucleoprotein mAb demonstrated strong expression of this protein in the cytoplasm of pulmonary alveolar epithelial cells (Fig. 4C). Tissue damage was primarily confined to lung tissues. However, there was mild perivascular mononuclear cell infiltration in the cerebral cortex from the mice treated with zanamivir alone but not in those from mice treated by both zanamivir and immunomodulators, whereas focal dense mononuclear cell infiltration in the cerebral cortex was observed in brain tissues taken from the untreated mice (Fig. S2A). Reactive lymphoid cells that appeared paler in staining were found in spleens obtained from zanamivir-treated and PBS control mice, in which reactive lymphoid cells were present along with frequent apoptotic bodies with prominent nuclear fragmentation, but not in those collected from mice treated with zanamivir and immunomodulators (Fig. S2B). Nevertheless, no significant pathologic changes or tissue damages could be detected in liver (Fig. S2C) and kidney (Fig. S2D) from all mice.

As shown in Fig. S3A, 12 surviving mice with undetectable viral load in lung tissues at day 21 after viral challenge also had a neutralizing antibody titer of 80. Western blot confirmed that the neutralizing antibody reacted specifically with baculovirus-expressed nucleoprotein and hemagglutinin of H5N1 (Fig. S3B). Interestingly, two surviving mice treated with triple therapy still had a detectable low viral load and a neutralizing antibody titer of 40. Compared with the zanamivir-treated group whose TCID₅₀ titer in the TPL was below our detectable limit, the triple therapy group had a TCID₅₀ titer of 5.1 × 10² ± 4.9 × 10², which was still 2.5 log below the titer of 2.7 × 10⁵ ± 2.0 × 10⁵ in the PBS control group (Fig. 2A). This finding is not completely unexpected because the immunomodulators may still have some degree of immunosuppression that is not clinically apparent. Consistent with these findings, these two mice [zanamivir + celecoxib + mesalazine (2)], together with the surviving mouse from the zanamivir-treated group, also had inflammatory infiltrate in their alveoli on histologic examination, whereas no significant inflammation was observed in the other surviving mice [zanamivir + celecoxib, zanamivir+ mesalazine, and zanamivir + celecoxib + mesalazine (6)], which was similar to those found in normal mice (Fig. 4B and Fig. S3C).

Discussion

There is an urgent need to find an effective treatment strategy against H5N1 infection in humans because of the substantial mortality associated with this virus. Although oseltamivir is highly effective in mouse models, humans treated with this drug still exhibited high fatality, which can be attributed to delayed initiation of therapy. Many antiviral treatment studies of mouse models infected by H5N1 used an inoculum of ≈10 LD₅₀ of H5N1. Good treatment results were obtained if the antiviral was started 4 h before, soon after, or within 36 h after inoculation (8, 9). Only a few studies showed good results when the antiviral treatment was started after 36 h. However, in those series, either a low viral inoculum was used or a duck H5N1 virus adapted to mice was used instead of a human virus for inoculation (10–12). Thus the pathophysiologic status of the infected mice in those studies could be quite different from the real clinical situation when patients often do not enter the hospital until 2–4 days after the onset of symptoms, when the viral load in respiratory secretions is already high. The high inoculum and delayed therapy in the presently reported mouse model provided a more realistic simulation for testing various forms of therapy. To avoid the confounding effects of poor oral bioavailability of oseltamivir in sick mice and the known risk of emergence of oseltamivir resistance during therapy, i.p. zanamivir was used. However, as in the case of oseltamivir, 87% of the mice died when the zanamivir treatment was delayed for 48 h, although the survival time was insignificantly prolonged. Our animal model provided an ideal situation for testing combination therapy with immunomodulators that had no antiviral effects or any significant effect on survival if used alone.

Our study showed that even if the viral replication had been suppressed in the mice treated with antiviral, levels of cytokines and chemokines were still similar to the untreated mice, which were significantly higher than those in the mice receiving combination therapy. This finding suggests that once the viral infection has triggered the cytokine storm, even if viral replication is suppressed by antiviral therapy, the proinflammatory cytokines and chemokines will continue to drive the immunopathologic progression. Previous studies showed that antiinflammatory dose of steroid was ineffective in mice in the absence of antiviral treatment (7) and was associated with significant side effects in human infected by the H5N1 virus without improving survival (5, 6). Therefore other immunomodulators have to be considered.

Because COX-2 knockout mice had significantly better survival after challenge with mouse-adapted influenza A H3N2 virus than wild-type BALB/c mice (13), i.p. celecoxib was chosen in this study. Sulfasalazine and related compounds such as mesalazine and 5-aminosalicylic acid are also chosen in this study because they are highly active in alimentary tract epithelial cells and are commonly used in the treatment of inflammatory bowel diseases. They have diverse effects on the immune system including inhibition of lipoxygenase and COX pathways, which decrease proinflammatory cytokines and eicosanoids, and therefore decrease the activation of inflammatory cells such as macrophages and neutrophils. In addition, sulfasalazine and 5-aminosalicylic acid inhibit NF-κB activation and promote the synthesis of phosphatidic acid. Both actions inhibit the potent stimulatory effects of ceramides on apoptosis (14, 15). The combined actions of mesalazine (the effective moiety of sulfasalazine) and celecoxib have synergistic effects in counteracting virus-induced cytokine dysregulation and apoptosis. Both drugs are relatively inexpensive, currently used in humans, not known to cause immunosuppression, and relatively free from adverse drug interactions or major side effects with short-term use.

The main target of action of the fibrates such as gemfibrozil is peroxisome proliferators-activated receptor α (PPARα). PPARs are members of the nuclear receptor superfamily that affects the lipid and glucose metabolism and inflammatory responses. PPARα activation inhibits NF-κB, COX-2 activity, and production of proinflammatory cytokines such as IL-6 and TNF-α (16). Therefore, activation of the PPARα by gemfibrozil is expected to damp down the excessive inflammatory response. Budd et al. (17) demonstrated that gemfibrozil improved survival of mice infected by influenza A/H2N2 virus from 26% (controls) to 52% (treated). However, no improved survival was noted when the hypervirulent H5N1 virus was used in this study. This discrepancy could be related to the different pathophysiology between H2N2 and H5N1 viruses or the relatively weak agonistic activities of gemfibrozil on PPARα.

The association between higher levels of PGE₂ and mice survival is compatible with the known immunologic profiles of human and experimental H5N1 infection. Among other cytokines and chemokines, severe H5N1 infections are associated with increased RANTES and MIP-1, and their synthesis is inhibited by PGE₂. Our triple therapy also showed a reduction in MIP-1 levels without suppressing PGE₂. PGE₂ has antiinflammatory and antiapoptotic properties, both of which may play a beneficial role in preventing excessive tissue and cellular damage. Previous reports showed that COX-2^−/− knockout mice had a significantly lower mortality, lesser degree of inflammatory cell infiltrates in the lungs, and lower levels of proinflammatory cytokines (TNFα, IL-1β, IFN-γ, IL-6) in the TPL as compared with wild-type and/or COX-1^−/− knockout mice after infection by influenza A/H3N2 virus (13). But the PGE₂ levels in the TPL and the viral load in the lungs were significantly higher in COX-2^−/− mice. Our findings of lower leukotrienes and higher PGE₂ levels in the TPL in mice treated by combination therapy is compatible with the above findings. Although mesalazine or celecoxib have not been shown to cause immunosuppression, two surviving mice after triple therapy still had a low, but detectable, viral load and a low level of neutralizing antibody. This finding is not unexpected because the same immunologic factors causing tissue damage during the mounting of the immune response may also be critical for viral clearance (18). The combination of mesalazine and celecoxib may cause mild subclinical immunosuppression. IL-1 was speculated to be protective because infected IL-1 receptor knockout mice showed increased morbidity, mortality, lung viral titer, and inflammatory infiltrate when infected with a low-lethality HK/486 virus (19). In this study, mice treated by triple therapy had improved survival without significant suppression of IL-1 in TPL despite the use of a hypervirulent virus.

H5N1-infected patients who succumbed often had persistently high levels of serum proinflammatory cytokines and chemokines (3, 5). Therefore, the pathogenesis was initially attributed to virus-induced cytokine storm. However, studies with knockout mice deficient in TNF-α, TNF receptor 1, TNF receptor 2, IL-6, chemokine (C-C motif) ligand 2, MIP-1α, and IL-1R (7) did not confer better survival after viral challenge when antivirals were not given. Moreover, recent studies showed that the levels of serum proinflammatory cytokines and chemokines correlated closely with the viral load (5). Therefore, the pathogenesis should involve the interplay between a rising viral load and the resulting proinflammatory response. An effective therapy should consist of combinations of an effective antiviral agent and immunomodulatory agents to control viral load and cytokine storm, respectively. This scenario is especially true if patients present late in the course of influenza, when local and systemic proinflammatory cascade are fully activated.

Postmortem examination of patients who succumbed to H5N1 infection often showed severe lymphopenia and lymphoid atrophy or necrosis in the spleen and other lymphoid tissues (1, 3). Our study also showed that both CD4⁺ and CD8⁺ T lymphocytes were significantly decreased in antiviral-treated and untreated mice during disease progression. However, unlike the use of the steroid or other immunosuppressants, the use of celecoxib and mesalazine with zanamivir maintains significantly higher levels of CD4⁺ and CD8⁺ T lymphocytes at days 6 and 8 postchallenge. Histopathologic examination also showed that reactive lymphoid cells with frequent apoptotic bodies were found in spleens obtained from zanamivir-treated and untreated mice, but were infrequent in spleens from mice treated with zanamivir and immunomodulators. Thus, the antiapoptotic effects of celecoxib plus mesalazine may play a role in averting this type of damage.

Our results provide a sound theoretical and experimental basis for further studies on the role of triple therapy. An antiviral and the combined use of celecoxib and mesalazine may cause synergistic reduction in proinflammatory cytokines, chemokines, and leukotrienes via different pathways. These inhibitory activities, together with the antiapoptotic activities of the aminosalicylates, reduce cell death and tissue damage in the host (20). Apoptosis in pulmonary alveoli and lymphoid tissues leading to lymphopenia are prominent pathological features in patients dying of H5N1 infection. The concomitant use of an effective antiviral is essential, not only to limit the extent of viral replication (which drives the cytokine dysfunction) from natural infection, but also to counteract the possible increase in viral load after COX-2 inhibition. We suggest that H5N1 avian influenza could be treated with an effective antiviral like i.v. zanamivir, in conjunction with immunomodulating drugs like celocoxib and mesalazine to control the symptoms associated with cytokine storm. Triple therapy offers some hope for surviving the devastating consequences associated with a pandemic influenza outbreak.

Materials and Methods

Animal Model and Viral Challenge.

BALB/c female mice, 5–7 weeks old, were kept in biosafety level-3 housing and given access to standard pellet feed and water ad libitum. All experimental protocols followed the standard operating procedures of the approved biosafety level-3 animal facilities and were approved by the Animal Ethics Committee (21). Aliquots of stocks of influenza A virus strain A/Vietnam/1194/04 were grown in embryonated eggs. Virus-containing allantoic fluid was harvested and stored in aliquots at −70°C. The LD₅₀ was determined in mice after serial dilution of the stock. One thousand LD₅₀ were used for viral challenge in all of the experiments. Infection was established by intranasal inoculation of mice anesthetized by isoflurane.

Antiviral and Immunomodulatory Treatments.

Antiviral and immunomodulators were administered by the i.p. route using 0.5-ml, 29-gauge ultrafine needle insulin syringes. The administered dosage for each agent followed protocols as described (18, 22–25). Control mice were given PBS i.p. on the same days (Table 1). Survival, body weight, and general conditions were monitored for 21 days or until death.

Table 1.

Treatment regimens containing zanamivir, celecoxib, mesalazine, and gemfibrozil used alone or in combination for infected mice

Group	Treatment regimen	No.
1	3 mg zanamivir in PBS i.p. once every 12 h × 8 days*	5
2	2 mg celecoxib in 10%DMSO/PBS i.p. once per day × 8 days*	5
3	1 mg mesalazine in ddH2O i.p. once per day × 8 days*	5
4	1 mg gemfibrozil in propylene glycol i.p. once per day × 8 days*	5
5	2 mg celecoxib + 1 mg mesalazine i.p. once per day × 8 days*	5
6	2 mg celecoxib + 1 mg gemfibrozil once per day × 8 days*	5
7	PBS i.p. once per day × 8 days*	5
8	3 mg zanamivir i.p. once every 12 h × 6 days†	33‡
9	3 mg zanamivir + 2 mg celecoxib i.p. × 6 days†	10
10	3 mg zanamivir + 1 mg mesalazine i.p. × 6 days†	10
11	3 mg zanamivir + 2 mg celecoxib + 1 mg mesalazine i.p. × 6 days†	33‡
12	PBS i.p. once per day × 6 days	33‡

BALB/c mice (female, ages 5–7 weeks) were intranasally challenged with 1,000 LD₅₀ of H5N1 virus strain A/Vietnam/1194/04.

*The treatments started 4 h postchallenge.

^†The treatments started 2 days postchallenge.

^‡ Experiments were conducted in triplicates of five mice for each group. Furthermore, six mice in each of these groups were killed on days 4, 6, and 8 postchallenge, while all survived mice were killed on day 21 postchallenge. Blood, TPL, lung, brain, kidney, liver, and spleen were collected from these mice.

Experiments were conducted in duplicates or triplicates of five mice for each group of treated or control mice. Six mice in each of groups (groups 8, 11, and 12 in Table 1) were killed on days 4, 6, and 8 postchallenge, respectively. Blood, TPL, lung, brain, kidney, liver, and spleen tissue samples were collected from these mice, normal uninfected mice, and the surviving mice for histopathologic, immunologic, and virologic assays.

Virologic Tests.

Titers of released virus in TPL were determined by TCID₅₀ as described, whereas viral RNA in lung tissues was quantified by real-time RT-PCR (26, 27). Briefly, total RNA in lysed lung tissues was extracted by using a RNeasy Mini kit (Qiagen) and reverse-transcribed to cDNA by using applied SuperScript II Reverse Transcriptase (Invitrogen). Viral nucleoprotein gene and internal control β-actin gene were measured by the SYBR Green Mx3000 Real-Time PCR System (Stratagene), using primers NP-forward, 5′-GAC CAG GAG TGG AGG AAA CA-3′; NP-reverse, 5′-CGG CCA TAA TGG TCA CTC TT-3′; β-actin-forward, 5′- CGT ACC ACT GGC ATC GTG AT-5′; and β-actin-reverse, 5′-GTG TTG GCG TAC AGG TCT TTG-3′.

ELISA.

Proinflammatory cytokines and chemokines IL-1, IL-6, IFN-γ, TNF-α (BD Biosciences), PGE₂, MIP-1β (R&D Systems), leukotriene (GE Healthcare), and lung injury indicator albumin (Bethyl Laboratories) in TPL and serum samples were tested by ELISA using the protocol as described (3) with modifications according to the instructions of the kit suppliers.

Elastase Activity Assay.

Elastase activity in TPL was measured by the addition of the elastase-specific chromogenic substrate N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide (Sigma) at a final concentration of 1 mM. After 30 min at room temperature, the change in optical density at a wavelength of 405 nm was measured.

Neutralization Assay.

Neutralizing antibody levels in mice sera were determined by neutralization assay by using the same virus strain for challenge in MDCK cells as described (3).

Western Blot.

Influenza A viral proteins NP from H5N1 strain A/Indonesia/5/2005, HA1 from H5N1 strain A/Vietnam/1203/2004 (Immune Technology), and HA2 from strain A/Vietnam/1194/04, which was expressed in baculovirus vector (BD Bioscience), were separated in 12% SDS/PAGE gel and then electroblotted onto PVDF membrane. The membranes were incubated with mouse sera at 1/200 dilution, washed, and then incubated with HRP-conjugated anti-mouse IgG mAb at a dilution of 1/1,000 (Abcam). The blots were detected by the ECL Western blotting detection system (Amersham Biosciences).

Flow Cytometry.

Blood cells from the mice were stained with fluorescein-labeled mAbs specific for mouse CD3, CD4, and CD8 (BD Pharmingen) and fixed with 4% p-formaldehyde overnight. The fixed blood cells were analyzed by flow cytometry (FACSCaliber; BD) as described (21).

Histopathologic Analysis.

The lung, brain, spleen, kidney, and liver tissues of challenged mice were immediately fixed in 10% buffered formalin and embedded in paraffin wax. Sections 4–6 μm in thickness were mounted on slides. Histopathologic changes were examined by H&E staining under a light microscope as described (3).

Immunohistochemical Assay.

Lung sections were stained as described (3) by using an antiinfluenza nucleoprotein mAb (HB65; ATCC) at 1:5,000 dilution, goat anti-mouse IgG, H and L chain-specific biotin conjugate (Calbiochem) at 1:2,000 dilution, and streptavidin/peroxidase complex reagent (Vector Laboratories).

Statistical Analysis.

Statistical analysis of survival time and rate were performed by the log rank Kaplan-Meier and χ² tests, respectively, whereas the others were calculated by Student's t test with Stata statistical software. Results were considered significant at P ≤ 0.05. The Cox proportional hazards model was used to estimate hazard ratios.