CD8 T cells utilize TNF-related apoptosis-inducing ligand (TRAIL) to control influenza virus infection

Abstract

Elimination of influenza virus infected cells during primary influenza virus infections is thought to be mediated by CD8⁺ T cells though perforin- and FasL-mediated mechanisms. However, recent studies suggest that CD8⁺ T cells can also utilize TNF-related apoptosis-inducing ligand (TRAIL) to kill virally-infected cells. Therefore, we herein examined the importance of TRAIL to influenza-specific CD8⁺ T cell immunity and to the control of influenza virus infections. Our results show that TRAIL deficiency increases influenza-associated morbidity and influenza virus titers, and that these changes in disease severity are coupled to decreased influenza-specific CD8⁺ T cell cytotoxicity in TRAIL^−/− mice—a decrease that occurs despite equivalent numbers of pulmonary influenza-specific CD8⁺ T cells. Further, TRAIL expression occurs selectively on influenza-specific CD8⁺ T cells, and high TRAIL receptor (DR5) expression occurs selectively on influenza-virus-infected pulmonary epithelial cells. Finally, we show that adoptive transfer of TRAIL^+/+ but not TRAIL^−/− CD8⁺ effector T cells alters the mortality associated with lethal dose influenza virus infections. Collectively, our results suggest that TRAIL is an important component of immunity to influenza infections, and TRAIL deficiency decreases CD8⁺ T cell-mediated cytotoxicity leading to more severe influenza infections.

Introduction

Primary infection with influenza virus results in a localized pulmonary infection and inflammation, and elicits an influenza-specific CD8⁺ T cell immune response that is necessary for viral clearance (1–7). These CD8⁺ T cells are thought to control virus infections by killing influenza-infected pulmonary cells using perforin- and Fas-dependent mechanisms (3). When Fas^−/− or perforin^−/− mice were infected with influenza virus in the absence of CD4⁺ T cells, viral titers persisted for an additional 3 days beyond controls but virus was eventually cleared (3). This result suggests that in the absence of one death-inducing pathway influenza-specific CD8⁺ T cells will compensate and utilize other cytotoxic mechanisms to eliminate influenza virus infected cells. In contrast, when perforin^−/− bone marrow reconstituted Fas^−/− mice were likewise infected with influenza virus, ~70% of the mice had sustained long term (i.e. 14 days p.i.) high viral titers (3) suggesting that influenza-specific CD8⁺ T cell cytotoxicity requires access to either the perforin or Fas cytotoxic pathways to effectively control influenza virus infections. Interestingly however, ~30% of the above perforin^−/−Fas^−/− mice were able to reduce pulmonary influenza virus titers leading to the idea that another cytotoxicity pathway could be involved in viral elimination (3).

CD8⁺ T cells have recently be described to use the TNF-related apoptosis-inducing ligand (TRAIL) pathway, in addition to the Fas:FasL and perforin/granzyme (lytic granule) pathways, to kill target cells (8). TRAIL has classically been studied in tumor immunology settings, where it selectively induces apoptosis in transformed cells while leaving non-transformed cells unaffected (9, 10). Beyond a role in tumor surveillance, TRAIL-based immunity is also a component of the immune response during viral infections, including responses to CMV, HIV, and RSV (11–13). Moreover, a previous study has shown that the expression of mRNA for TRAIL and its receptor DR5 (TRAIL-R2) are increased in the lungs during influenza virus infections, that TRAIL is expressed by T cells in the lungs of influenza virus-infected mice, and that clearance of influenza virus is delayed by administering a blocking anti-TRAIL mAb during primary infections (14). While these results suggest a role for TRAIL in immunity to influenza virus infections, it remains unknown if the expression of TRAIL by T cells during influenza infections is limited to just influenza-specific T cells, if influenza-specific CD8⁺ T cells utilize TRAIL to kill influenza-infected cells and control virus infection, and how TRAIL deficiency alters the course and magnitude of influenza virus infections. Therefore, we utilized TRAIL^+/+ and TRAIL^−/− mice to determine the contribution of TRAIL to the influenza-specific CD8⁺ T cell immune response during primary influenza virus infections. Our results confirm a role for TRAIL in the primary immune response to influenza virus infection, and demonstrate that TRAIL-mediated apoptosis is a third mechanism which influenza-specific CD8⁺ T cells can use to eliminate influenza-infected cells and drive recovery from influenza.

Materials and Methods

Mice, virus, peptides, and infections

C57Bl/6 (TRAIL^+/+; H-2^b) mice were purchased from the National Cancer Institute (Frederick, MD). C57Bl/6 TRAIL-deficient mice (TRAIL^−/−; H-2^b) were obtained from Amgen (Seattle, WA) (15) and C57BL/6 DR5-deficient mice (DR5^−/−, H-2b) were obtained from WS El-Deiry (16). Knockout mice were bred in our own facility at the University of Iowa, according to UI IACUC guidelines. They are >10 generations backcrossed to C57Bl/6. All mice were used at 12–20 weeks of age, and all animal experiments followed approved IACUC protocols. The mouse-adapted influenza A virus A/PuertoRico/8/34 (PR8; H1N1) was grown in the allantoic fluid of 10 d old embryonated chicken eggs for 2 d at 37°C, as previously described (17, 18). Allantoic fluid was harvested and stored at −80°C. Groups of 24.5–27.5 g TRAIL^+/+ and TRAIL^−/− mice were given a 500EIU dose of mouse-adapted PR8 virus in Iscove's media intranasally (i.n.) following anesthesia with halothane. The peptides used in this study, NP₃₆₆ (ASNENMETM) and PA₂₂₄ (SSLENFRAYV) were purchased from Biosynthesis Inc. (Lewisville, TX), and are derived from the amino acid sequence of A/PR/8/34 nucleoprotein (NP) or acid polymerase (PA), respectively (19–21).

Lung Virus Titer

Pulmonary viral titers were determined via endpoint dilution assay and expressed as Tissue Culture Infections Dose₅₀ (TCID₅₀). Briefly, 10-fold dilutions of homogenized and clarified lung from influenza virus infected mice were mixed with 10⁵ MDCK cells in DMEM. After 24 h incubation at 37°C, the inoculum was removed and DMEM media containing 0.0002% L-1-(tosylamido-2-phenyl)ethyl chloromethyl ketone (TPCK)-treated trypsin (Worthington Diagnostics, Freehold, NJ) and penicillin (100U/ml)/streptomycin (100mg/ml) was added to each well. After 3 d incubation at 37°C in a humidified atmosphere of 5% CO₂, supernatants were mixed with an equal volume of 0.5% chicken RBC, the agglutination pattern read, and the TCID₅₀ values calculated.

Quantitative RT-PCR

Total RNA was harvested from homogenized lungs with TRIzol reagent (Invitrogen, Carlsbad, CA). Total RNA (2 mg) was reverse-transcribed using Superscript II. The quantitative PCR primer/probe sets for mouse TRAIL, DR5, Fas, FasL, perforin, granzyme B, and rRNA were purchased from PE Applied Biosystems (Foster City, CA). 250 ng of cDNA was used as a template for TaqMan assays for all transcripts and the internal rRNA control. The TaqMan PCR reaction was carried out as described previously (22).

Cytotoxicity Assays

Splenocytes from wildtype DR5^+/+ and DR5^−/− C57Bl/6 mice (16) were resuspended in NycoPrep 1.077A (Axis-Shield; Norton, MA) and then purified according to the manufacturer's instructions. NycoPrep-purified splenic mononuclear cells (10⁷/ml) were labeled with either 2 µM CFSE (Invitrogen; Eugene, OR) at 37°C for 10 min or 2 µM PKH-26 (Sigma; St. Louis, MO) at room temperature for 5 min. After labeling, residual non-cell-associated CFSE and PKH-26 were neutralized by adding an equal volume of fetal calf serum to the cell suspension. CFSE-labeled splenic mononuclear cells (10⁷/ml) were pulsed with 10 µM PA₂₂₄ and NP₃₆₆ peptide for 1 h at 37°C. PKH-26⁺ splenic mononuclear cells (10⁷/ml) were similarly incubated without peptide for 1 h at 37°C. The cells were then washed and mixed at a 1:1 ratio, and 10⁷ cells (i.e., 5 × 10⁶ CFSE⁺, 5 × 10⁶ PKH-26⁺ cells) were adoptively transferred i.v. into influenza-virus-infected TRAIL^+/+ or TRAIL^−/− mice. After 8 h, the lungs were removed, digested, and analyzed by flow cytometry as previously described (18).

Flow-Cytometry Analysis

Surface Labeling

Isolated lung cells (10⁶) were stained with: PE, PerCP-CY5.5, or APC-conjugated anti-mouse CD8α (53-6.7; BD Biosciences, San Jose, CA); biotinylated anti-mouse CD178/FasL (MFL3; eBioscience, San Diego, CA); or biotinylated anti-mouse TRAIL (N2B2; eBioscience). Cells stained with biotinylated mAb were subsequently incubated with Strepavidin-PerCP, Strepavidin-PE, or Strepavidin-APC (BD Bioscience). Stained cells were fixed and erythrocytes lysed with FACS lysing solution (BD Biosciences), and subsequently analyzed on a FACSCalibur flow cytometer. NP₃₆₆ and PA₂₂₄ tetramers were obtained from the National Institute of Allergy and Infectious Disease MHC Tetramer Core Facility (Germantown, MD).

Pulmonary Epithelial Staining

Isolated lung cells (10⁶) were stained with the FITC-conjugated anti-mouse T1α/Podoplanin (8F11; MBL, Woburn, MA) or isotype control, and PE-conjugated anti-mouse DR5 (MD5-1; eBioscience) or isotype control. Subsequently, the cells were fixed, permeabilized, and stained with biotinylated anti-NP (H16-L10-4R5; a kind gift from Walter Gerhard, Wistar Institute, University of Pennsylvania). Biotinylated antibody was subsequently revealed with PerCP-CY5.5.

Intracellular Staining

Granzyme B. Isolated lung cells (10⁶) were surfaced stained with PerCP-CY5.5-conjugated anti-mouse CD8α. Subsequently, the cells were fixed, permeablized, and stained with the PE-conjugated anti-human Granzyme B mAb (GB11; Invitrogen), or isotype control (23). IFN-γ Cells from mice infected with influenza were cultured at 2 × 10⁶ cells/well in the presence of 1 µM of influenza peptides or media control, FITC-conjugated anti-CD107a (1D45; eBioscience) or isotype control, 400 U/ml recombinant human IL-2, and 1 µg/ml brefeldin A. After 6 h, cells were harvested, stained with PE-conjugated rat anti-mouse CD8α, fixed, permeablized, and stained with APC-conjugated rat anti-mouse IFN-γ (XMG1.2; eBioscience) or isotype control (24).

CD8⁺ T cell Adoptive Transfer

Groups of 24.5–27.5g TRAIL^+/+ (CD45.2) and TRAIL^−/− (CD45.2) mice were given a 500EIU dose of mouse-adapted A/PR/8/34 virus in Iscove's media intranasally (i.n.) following anesthesia with halothane. On day 8 p.i., single-cell suspensions of pulmonary cells from the infected TRAIL^+/+ or TRAIL^−/− mice were incubated with anti-CD8α microbeads and CD8⁺ cells purified according to the manufacturer's instructions (Miltenyi Biotec, Auburn, CA). The purified CD8⁺ cell were then transferred intraveneously into CD45.1 TRAIL^+/+ mice (obtained from Dr. Robert Cook, University of Iowa) that had previously been infected with a lethal dose (2200EIU) of mouse-adapted A/PR/8/34. Cell transfer into lethally infected mice occurred on d 5 p.i. Overall morbidity and mortality of the mice was monitored to 21 days after lethal infection.

Statistical analysis

For each analysis, normal distribution of data was first verified. To assess the difference between two sets of data with normal distribution, statistical significance was assessed using an unpaired, one-tailed t-test or a paired t-test for control and experimental data groups that could be paired. If normality test failed, Mann-Whitney Rank Sum tests were completed to compare data sets. To assess the differences among multiple sets of data with normal distribution, statistical significance was assessed using an ANOVA analysis of the data sets. If normality test failed, Kruskal-Wallis One Way Analysis of Variance on Ranks test was used to determine overall significance with subsequent pair wise comparisons completed using Dunn’s Method. To determine differences in survival and viral clearance, Kaplan-Meier Survival Analysis: Log-Rank tests were run to determine significant differences between data sets. When appropriate, subsequent pair wise multiple comparisons were completed using the Holm-Sidak method. Differences were considered to be statistically significant at p ≤ 0.05.

Results

TRAIL-deficient mice display increased morbidity and influenza loads during influenza infection

To rigorously investigate the role that TRAIL plays in the regulation of influenza virus infections, we initially determined the impact of TRAIL deficiency on the severity of influenza virus infections. As shown in Figure 1A, TRAIL ^−/− mice demonstrated significant (p < 0.01) weight loss (i.e. morbidity) relative to wildtype C57Bl/6 (TRAIL^+/+) controls following infection with a low dose of influenza virus. This increase in disease severity correlated with increased pulmonary viral titers (Fig 1B). Specifically, while the amount of virus in the lungs of TRAIL^+/+ animals was reduced by ~1 log between days 4–6 p.i., TRAIL^−/− mice showed little change in the amount of infectious virus present in their lungs. Furthermore, while TRAIL^−/− animals were able to eventually reduce pulmonary virus levels by day 8 post infection (p.i.), the number of TRAIL^−/− mice that had cleared virus below the limit of detection remained significantly reduced relative to TRAIL^+/+ animals. Together, these results suggest that the increased morbidity observed in TRAIL^−/− mice might, in part, be tied to an increased and sustained pulmonary viral burden.

FIGURE 1.

TRAIL deficiency correlates with increased disease severity after influenza virus infection. A, C57BL/6 TRAIL^+/+ (○) or TRAIL^−/− (▲) mice (n = 4 mice/group) were infected with influenza and weighed daily to assess morbidity. The values displayed represent the daily weight relative to the weight on day of infection (i.e. starting weight). *, p < 0.05, Mann-Whitney Rank Sum Test. No significant differences in mortality existed between the two groups. Data are representative of 2 separate experiments. B, Pulmonary virus titers were assessed by determining TCID₅₀ in MDCK cell cultures (as described in the Materials and Methods). *, p = 0.002, Mann-Whitney Rank Sum Test. Viral clearance was significantly different between the TRAIL^+/+ and TRAIL^−/− with a p = 0.036 as analyzed by a Kaplan-Meier Survival Analysis: Log-Rank.

Influenza-induced TRAIL and DR5 expression

Since our above results suggested that TRAIL played a significant role in the control and resolution of influenza virus infections, we next examined TRAIL and DR5 mRNA expression in influenza-infected lungs. The amount of TRAIL mRNA was increased in total lung homogenates from TRAIL^+/+ mice following i.n. influenza virus infection (as expected, no TRAIL mRNA was detected in the TRAIL^−/− mice; Figure 2). Moreover, DR5 mRNA expression was similarly upregulated in the lungs of both TRAIL^+/+ and TRAIL^−/− mice following i.n. influenza virus infection, indicating that the lack of TRAIL expression did not significantly affect DR5 mRNA expression in the TRAIL^−/− mice. Given that the upregulation of TRAIL and DR5 mRNA (starting at 6–8 days p.i.) in TRAIL^+/+ mice corresponds with the timing of increased disease in TRAIL^−/− mice (Fig 1A), these results further support the concept that TRAIL plays a major role in mediating the control and course of influenza virus infection. These results also largely confirm similar data recently described by Ishikawa and colleagues (14); however, unlike their results we do not observe significant increases in pulmonary TRAIL mRNA expression until 8 days p.i. (as opposed to 4 days p.i.). Furthermore, we observed a more rapid decrease in TRAIL and DR5 mRNA expression from their peak at 8–14 days p.i. These differences may be related to the difference in virus inoculum administered (500 EIU herein vs. 25 PFU), as well as the corresponding alterations in the inflammatory cytokines produced.

FIGURE 2.

Expression of effector molecule and receptor mRNA is equivalent in the lungs of TRAIL^+/+ and TRAIL^−/− mice during influenza virus infection. TRAIL, DR5, FasL, Fas, perforin, and Granzyme B mRNA expression in the lungs of TRAIL^+/+ (■) and TRAIL^−/− (●) mice were determined by quantitative RT-PCR on d 4, 6, 8, 14, 18, and 21 after infection. 18s rRNA was used to normalize the gene expression. Data are representative of 2 independent experiments.

To verify that the above changes in DR5 mRNA expression were reflected in alterations of DR5 protein expression, we next determined DR5 expression on pulmonary epithelial cells prior to and during an influenza virus infection. While there was a low amount of DR5 expressed on the surface of alveolar epithelial type I cells (i.e. T1α⁺ cells)(25) from uninfected mice (Fig 3A), overall DR5 expression was significantly increased following influenza virus infection. Moreover, the upregulation of DR5 by epithelial cells correlated with those cells that had been directly infected with influenza virus (i.e. NP⁺ cells), as NP⁺ epithelial cells expressed ~5x more DR5 relative to NP⁻ epithelial cells from the same lungs (Fig 3 A, B). These data suggest that influenza infection of pulmonary epithelial cells results in selective upregulation of DR5 on influenza-infected lung epithelial cells, potentially increasing their susceptibility to TRAIL-mediated lysis.

FIGURE 3.

Pulmonary expression of TRAIL-receptor (DR5) and TRAIL occur in an influenza-specific fashion. At d 4 p.i., lungs were harvested from TRAIL^+/+ mice and prepared into a single cell suspension. A, Isolated cells were stained with anti-T1α, anti-DR5, and anti-NP or respective isotype controls. The top histogram represents the basal DR5 expression in uninfected T1α-positive pulmonary cells (dashed line) relative to the DR5-isotype control (shaded histogram). The bottom histogram shows DR5 expression on T1α⁺/NP⁺ (solid line) and T1α⁺/NP⁻ (dashed line) pulmonary cells relative to the isotype control (shaded histogram). B, DR5 mean fluorescence intensity (MFI) on NP⁺ and NP⁻ T1α⁺ cells from the lungs of infected mice (n = 5). Data are representative of 3 individual experiments. C, On day 8 p.i. cells isolated from TRAIL^+/+ mice were stained with anti-CD8α, anti-CD3ε, NP₃₆₆ and PA₂₂₄ tetramers, and anti-TRAIL (open histograms) or isotype controls (shaded histograms). Data are representative of 5 mice from 2 experiments. Similar influenza-specific expression of TRAIL on CD8⁺ T cells was observed in the draining lymph nodes on day 6p.i (data not shown).

The greatest difference in morbidity between TRAIL^+/+ and TRAIL^−/− mice was seen after 6 days p.i., paralleling the increase in TRAIL and DR5 expression. These kinetics are similar to the described kinetics of influenza-specific CD8⁺ T cell recruitment into the lungs (18, 23). Since CD8⁺ T cells mediate the clearance of influenza-infected cells (1–5), our results suggested that the increased disease severity and viral burden in TRAIL^−/− mice might be linked to altered pulmonary influenza-specific CD8⁺ T cells responses. Therefore, we next examined TRAIL expression on influenza-specific CD8⁺ T cells in the lungs during influenza infections, and found that TRAIL was indeed expressed by influenza-specific CD8⁺ T cells in the lungs of TRAIL^+/+ mice (Fig 3C). In contrast, the vast majority of non-influenza-specific CD8⁺ T cells within the lungs at the same time did not appear to express TRAIL. The small residual TRAIL⁺ shoulder in the NP₃₆₆ or PA₂₂₄ negative T cell populations is likely attributable to the remaining unstained immunodominant PA₂₂₄ or NP₃₆₆ specific cells, respectively, with potential minor contributions from the other 6 subdominant epitopes (19, 26). Similar to the correlation with influenza CD8⁺ T cell epitope specificity observed in the lungs, TRAIL was also selectively expressed on influenza-specific CD8⁺ T cells within the lung draining lymph nodes on day 6 p.i. (data not shown). Consistent with a previous report demonstrating TCR stimulus driven upregulation of TRAIL on naϊve and effector CD8 T cells (27), we also observed antigen-specific upregulation of TRAIL by naϊve influenza-specific CD8 T cells after 24 hours of in vitro culture (data not shown). Together the antigen-specific nature and location of TRAIL expression suggest that TRAIL expression by CD8⁺ T cells likely relates to their initial programming within the draining lymph nodes (27–30) and could be amplified by interactions with viral-peptide MHC I complexes in the lungs (27).

Influenza-specific CD8⁺ T cell response in TRAIL^−/− mice

The observed difference in morbidity between TRAIL^+/+ and TRAIL^−/− mice, combined with the selective expression of TRAIL on influenza-specific CD8⁺ T cells, suggests vigorous resolution of the infection requires the participation of TRAIL-expressing CD8⁺ T cells. The likely explanation for the increased morbidity in the TRAIL^−/− mice is that the influenza-specific CD8⁺ T cells are unable to kill influenza-infected cells, but it may also be possible that other factors contribute to the pathology, such as a reduction in the number of lung-infiltrating antigen-specific effector CD8⁺ T cells. Thus, we examined the magnitude and phenotype of the pulmonary CD8⁺ T cell response in influenza-infected TRAIL^+/+ and TRAIL^−/− mice. Interestingly, TRAIL deficiency did not alter the magnitude of the NP₃₆₆ or PA₂₂₄ influenza-specific CD8⁺ T cell response in the lungs (Fig 4A & B). The level of IFN-γ produced per cell was also similar between both influenza antigen-specific TRAIL^+/+ and TRAIL^−/− CD8⁺ T cells (TRAIL^+/+ NP₃₆₆ MFI= 235; TRAIL^−/− NP₃₆₆ MFI= 303; TRAIL^+/+ PA₂₂₄ MFI= 242 TRAIL^−/− PA₂₂₄ MFI=423; not significant as determined by Kruskal-Wallis One Way Analysis of Variance on Ranks). However, TRAIL deficiency did result in a significant reduction in influenza-specific CD8⁺ T cell-mediated cytotoxicity in vivo (Fig 4C). Indeed, while TRAIL^−/− mice have an equal in vivo E:T ratio to that of TRAIL^+/+ mice, the influenza-specific CD8⁺ T cells killed wildtype DR5^+/+ influenza peptide-pulsed targets with significantly reduced (~40%) efficiency. Further, when influenza peptide-pulsed DR5^−/− targets were adoptively transferred into either influenza infected TRAIL^−/− or TRAIL^+/+ hosts, killing was reduced ~60% relative to DR5^+/+ targets in TRAIL^+/+ animals.

FIGURE 4.

Despite similar CD8⁺ T cell responses, cytotoxicity is decreased in influenza virus-infected TRAIL^−/− animals. At d 8 p.i., lungs were harvested from C57BL/6 TRAIL^+/+ (□) or TRAIL^−/− (■)mice. A, Isolated cells from TRAIL^+/+ or TRAIL^−/− mice were stained with anti-CD8α, NP₃₆₆ tetramer or PA₂₂₄ tetramer, and anti-CD3ε and the number of CD8⁺tetramer⁺ T cells (mean ± SD) enumerated using total counts and flow cytometry. ANOVA analysis yielded no significant differences among the number of TRAIL^+/+ and TRAIL^−/− tetramer⁺ T cells or between the individual tetramers. Data are representative of 3 experiments. B, Pulmonary cells were incubated with NP₃₆₆ and PA₂₂₄ peptides (or control media); the frequency of antigen-specific T cells was measured by IFNγ ICS. Shown is the percentage of IFN-γ⁺ of CD8⁺ cells. Equal numbers of IFN-γ⁺CD8 T cells were observed in both groups (data not shown). ANOVA analysis yielded no significant differences between the number or percentage of TRAIL^+/+ and TRAIL^−/− IFNγ⁺ T cells or between the two epitopes. Data are representative of 2 experiments. C, The pulmonary influenza-specific CD8⁺ T cell response in TRAIL^+/+ (○) or TRAIL^−/− (●) mice was measured by in vivo cytotoxicity assay on d 8 p.i. Target cells were purified from DR5^+/+ and DR5^−/− were used as indicated. Targets from DR5^+/+ were verified to be DR5⁺ by flow cytometry (data not shown). Percent influenza-specific killing was calculated by comparing unpulsed target lysis to influenza-peptide pulsed target lysis. Target frequencies were normalized to ratios harvested from transfers into naϊve mice. *, p = 0.029 determined using a 24 paired t-test.

The reduced cytotoxicity of TRAIL^−/− CD8⁺ T cells or during transfer of DR5^−/− targets was intriguing given the increased viral load and disease severity observed in TRAIL^−/− mice (Fig 1). Together our results suggest that between 40–60% of influenza-specific CD8 T cell mediated cytotoxicity within the lungs on day 8 p.i. may be TRAIL:DR5 dependent. Yet disruption of a single cytotoxicity pathway was not expected to have major pathological consequences, based upon the redundant roles of the Fas/FasL and perforin/Granzyme pathways in mediating control of influenza virus infections (3). In those studies, mice deficient in either Fas or perforin showed only marginal changes in pulmonary viral load. It took a deficiency in both perforin⁺ T cells and Fas⁺ targets cells to sustain viral titers until day 14 post infection - a time point when virus has normally been eliminated from the lungs. Therefore, we next measured the expression of Fas, FasL, perforin, and Granzyme B mRNA in the lungs, as well as FasL and Granzyme B protein expression and degranulation potential in TRAIL^+/+ and TRAIL^−/− influenza-specific CD8⁺ T cells found within the lungs. Fas, FasL, Granzyme B, and perforin mRNA were all upregulated to similar levels in both TRAIL^+/+ and TRAIL^−/− mice peaking at day 8 p.i. (Fig 2). Further, neither NP₃₆₆ or PA₂₂₄ specific TRAIL^−/− CD8⁺ T cells had altered expression of FasL or Granzyme B or changes in the ability to degranulate (i.e. surface CD107a expression) relative to TRAIL^+/+ controls (Fig 5). Therefore, these results collectively suggest TRAIL:DR5 interactions may play more prominent roles in CD8⁺ T cell-mediated clearance of influenza virus infected cells than has previously been appreciated.

FIGURE 5.

TRAIL^−/− and TRAIL^+/+ influenza-specific CD8⁺ T cells have similar granzyme B and FasL expression as well as similar degranulation. At d 8 p.i., lungs were harvested from C57BL/6 TRAIL^+/+ or TRAIL^−/− mice. A, Isolated cells were stained with anti-CD8α, NP₃₆₆ tetramer, PA₂₂₄ tetramer, anti-granzyme B or isotype control, or anti-FasL or isotype control mAb. Upper panels show Granzyme B (left) or FasL expression (right) on CD8⁺NP₃₆₆ ⁺ cells from C57BL/6 TRAIL^+/+ or TRAIL^−/− mice. Lower panels show Granzyme B (left) or FasL expression (right) on CD8⁺PA₂₂₄ ⁺ cells from TRAIL^+/+ or TRAIL^−/− mice. Gray histograms represent isotype control staining. Histograms are representative of 5 mice and 2 separate experiments. B, Isolated cells were incubated with NP₃₆₆ and PA₂₂₄ (or control media), BFA, and anti-CD107a for 5 h. After incubation, the cells were stained with anti-CD8 and anti-IFN-γ. Histograms represent the CD107a expression on CD8⁺IFN-γ⁺ cells from TRAIL^+/+ or TRAIL^−/− mice. Histograms are representative of 5 mice and 2 separate experiments.

TRAIL^+/+ CD8 T cells protect from ongoing lethal influenza virus infections

Since our results suggested that influenza-specific CD8⁺ T cells utilize TRAIL to eliminate virally-infected pulmonary epithelial cells and therein control virus infections, we tested the ability of TRAIL^+/+ and TRAIL^−/− influenza-specific effector CD8⁺ T cells to control and resolve an ongoing lethal-dose influenza virus infection in TRAIL^+/+ mice. During such infections, endogenous pulmonary influenza-specific CD8⁺ T cells and virus control are limited due to a previously described elimination of effector CD8⁺ T cells during their development in the lymph nodes (18). However, this lethality can be overcome when normal T cell numbers are restored to the lungs (18). When we intravenously adoptively transferred TRAIL^+/+ pulmonary effector CD8⁺ T cells 5 days p.i into lethal dose influenza infected mice, 83.3% of the mice survived and recovered from the high dose influenza infection. In contrast, while the donor TRAIL^−/− effector T cells migrated into the lungs at equivalent numbers to TRAIL^+/+ effector T cells (data not shown), the TRAIL^−/− effector T cells were only able to protect 33.3% of the lethally-infected mice, a percentage that was not statistically different from the non-T cell transferred controls. Since no differences in IFNγ, FasL, Granzyme B, or degranulation were observed in the NP₃₆₆ and PA₂₂₄ effector CD8 T cell populations (Fig 5) and these T cells arrived in the lungs in equivalent numbers upon adoptive transfer, our results suggest that the TRAIL expression or deficiency alone is responsible for the differential ability to protect these mice from lethal-dose influenza virus infections.

Discussion

TRAIL-expressing CD8⁺ T cells, NK cells, and plasmacytoid DC have all been implicated in cytotoxicity and control of viral infections (31–33). The results presented herein suggest that the differences observed during influenza virus infection of TRAIL^+/+ and TRAIL^−/− mice are predominately due to altered CD8⁺ T cells responses. First, while some minor increases in morbidity were observed within the first 3 days following influenza virus infection (i.e. the window of time normally ascribed to innate [NK/pDC] control of infection), significant increases were not seen until after influenza-specific CD8⁺ T cells have arrived in the lungs (i.e. ~day 4+ p.i.) (18, 23, 34). Further, the kinetics of TRAIL and DR5 expression within the lungs correspond with the appearance of influenza-specific CD8⁺ T cells in the lungs, TRAIL expression by pulmonary CD8⁺ T cells appears to be directly linked to influenza-virus specificity, and adoptive transfer of TRAIL^+/+, but not TRAIL^−./− T cells, can mediate protection from ongoing lethal dose influenza virus infections. Finally, our in vivo cytotoxicity experiments directly show that TRAIL deficiency on T cells or DR5 deficiency on target cells results in significantly (p < 0.029) reduced CD8⁺ T cell cytotoxicity of influenza peptide pulsed target cells despite the presence of equal numbers of pulmonary influenza-specific CD8⁺ T cells and identical levels of IFNγ, FasL, Granzyme B, and degranulation by influenza-specific TRAIL^+/+ and TRAIL^−/− CD8 effectors.

Of note, our results show that TRAIL expression appears to selectively correlate only with those pulmonary CD8⁺ T cells that are specific for influenza. Further, upregulation of DR5, the receptor for TRAIL, to high levels on pulmonary epithelial cells is linked to direct infection of the cells with influenza virus. Together these results suggest that elimination of virus-infected cells by influenza-specific CD8⁺ T cells could be specific not only at the level of recognition of virus peptide/MHC I complexes but also by the high level of DR5 expression. In this manner high expression of DR5 may permit T cell-mediated elimination of virally infected cells and allow survival of any surrounding non-infected epithelial cells or even pulmonary APC that carry viral peptide MHC complexes but have not upregulated DR5 due to direct infection - an idea that would be consistent with previously described selective TRAIL cytotoxicity within tumor systems (35). It is important to keep in mind that the amount of DR5 expressed on a cell’s surface is not the only point that determines TRAIL susceptibility. The events that are required for TRAIL-resistant cells to become susceptible to TRAIL are complicated and not well understood. Many viruses significantly alter host cell metabolism, such that it might be predicted that cells infected with viruses acquire sensitivity to death-inducing ligands (including TRAIL). Normal cells infected with RSV, human cytomegalovirus, or encephalomyocarditis virus become susceptible to TRAIL-mediated killing (11, 12, 36, 37), and we have also found the TRAIL-resistant human lung adenocarcinoma cell line, A549, can be sensitized to TRAIL following influenza virus infection (ELB, TSG, KLL unpublished data). While it is beyond the scope of this report, we are actively investigating the mechanism(s) that regulate TRAIL susceptibility in influenza virus-infected cells.

In the current study, TRAIL expression by CD8⁺ T cells correlates with reduced viral loads and disease severity; however, increased TRAIL expression may also lead to increased disease during some influenza virus infections (38). H5N1 influenza virus-infected human monocyte-derived macrophages express TRAIL at levels that are able to kill T cells - an outcome that could be inhibited by introduction of anti-TRAIL-R2 blocking antibodies (38). In this manner the TRAIL-expressing macrophages are thought to help drive the T cell lymphopenia observed with H5N1 influenza infections (38, 39). Influenza infection commonly induces the production of type I and II IFN as part of the innate immune response (40–42). Both types of IFN are potent inducers of TRAIL expression on many cells in the immune system (43, 44), and monocytes/Mϕ are exquisitely sensitive to IFN-induced TRAIL expression (45). Thus, while direct H5N1 influenza infection can induce TRAIL expression, the breadth and amount of TRAIL expressed by cells of the immune system may be further enhanced by IFN-mediated events. Therefore, our data and the results from the above studies would collectively suggest that TRAIL expression during influenza infections normally has a beneficial effect on viral control (i.e. CD8⁺ T cell-mediated elimination of infected pulmonary epithelial cells, etc), but that TRAIL may also serve to enhance the virulence of some influenza virus infections. The factors that regulate the beneficial versus deleterious effects, and possibly distinct cellular expression patterns of TRAIL during influenza infections, await further study.

Our observation that TRAIL^−/− mice do not significantly reduce viral titers as substantially from day 4 to 6 p.i. when compared to TRAIL^+/+ mice (Fig 1B) suggests that TRAIL-mediated cytotoxicity by CD8⁺ T cells may be more important than the Fas and perforin pathways of cytotoxicity during the early stages of influenza infections. This increased dependence upon TRAIL at these early times might relate to the low numbers of influenza-specific CD8⁺ T cells present (18, 23, 34), and hence low functional in vivo effector: target ratios in the lungs – an idea which would be consistent with TRAIL-dependent killing of some target cell lines in vitro (46). However, at later the stages of infection (i.e. days 6–8 p.i.) when the number of effector CD8⁺ T cells has significantly expanded (18, 23, 34), the loss of TRAIL may be compensated for by the other cytotoxicity pathways resulting in redundant and overlapping mechanisms of viral control and the 1 log reduction in pulmonary virus levels. Regardless, our results importantly show that a third pathway (i.e. TRAIL: DR5) of cytotoxicity is used along with the previously described Fas- and perforin-dependent killing pathways to eliminate and control influenza virus infection.

In conclusion, the results presented herein show that TRAIL plays a role in the regulation and control of influenza virus infections. Specifically, the early adaptive influenza-specific CD8⁺ T cell response appears to utilize TRAIL-mediated lysis of DR5⁺ (i.e. TRAIL receptor) influenza-infected cells in addition to FasL;Fas and perforin:granzyme dependent cytotoxicity pathways to control influenza virus infection. Further our results show that TRAIL and DR5 upregulation by CD8⁺ T cells and pulmonary epithelial cells is closely linked to either the antigen specificity of the T cells or the infection status of the epithelial cells, respectively. This suggests that TRAIL:DR5 specific interactions may partner with TCR:viral peptide-MHCI interactions to allow the targeted elimination of only influenza virus infected cells.

FIGURE 6.

Transfer of pulmonary CD8⁺ T cells from influenza-infected TRAIL^+/+ mice, but not TRAIL^−/− mice, reduces the mortality of lethal dose influenza-infected mice. At d 8 p.i., lungs were harvested from C57BL/6 TRAIL^+/+ or TRAIL^−/− mice. CD8⁺ T cells were isolated and transferred i.v. into mice that had been previously infected with a lethal dose of influenza virus (transfer made at day 5 p.i.). The values displayed represent the current percentage of mice surviving after receiving TRAIL^+/+ T cells, TRAIL^−/− T cells, or no transfer of cells. Data represent 2 pooled experiments (total mice no transfer n=9; total mice receiving TRAIL^−/− T cells n=9; total receiving TRAIL^+/+ T cells n = 6). No transfer vs. TRAIL^+/+ transfer, p = 0.00551; TRAIL^+/+ transfer vs. TRAIL^−/− transfer, p = 0.05; no transfer vs. TRAIL^−/− transfer, p = 0.517.

Abbreviations

TRAIL

TNF-related apoptosis-inducing ligand

DR5

death receptor 5 or TRAIL receptor 2