COINFECTION WITH STREPTOCOCCUS PNEUMONIAE NEGATIVELY MODULATES THE SIZE AND COMPOSITION OF THE ONGOING INFLUENZA-SPECIFIC CD8⁺ T CELL RESPONSE

Abstract

Infection with influenza A virus (IAV) can lead to increased susceptibility to subsequent bacterial infection, often with Streptococcus pneumoniae. Given the substantial modification of the lung environment that occurs following pathogen infection, there is significant potential for modulation of immune responses. Here we show that coinfection of mice with influenza virus followed by the non-invasive EF3030 strain of Streptococcus pneumoniae leads to a significant decrease in the virus specific CD8⁺ T cell response in the lung. Adoptive transfer studies suggest this reduction contributes to disease in coinfected animals. The reduced number of lung effector cells in coinfected animals was associated with increased death as well as a reduction in cytokine production in surviving cells. Further, cells that retained the ability to produce interferon γ exhibited a decreased potential for co-production of tumor necrosis factor α. Reduced cytokine production was directly correlated with a decrease in the level of mRNA. Negative regulation of cells in the mediastinal lymph node (MLN) was minimal compared to that present in the lung, supporting a model of negative regulation in the tissue harboring high pathogen burden. These results elucidate a new aspect of immune regulation as a result of entry of a coinfecting pathogen, modulation of ongoing adaptive immune responses in the lung. These findings reveal a novel dynamic interplay between concurrently infecting pathogens and the adaptive immune system.

INTRODUCTION

Influenza A virus (IAV) associated bacterial pneumonia is a significant cause of morbidity and mortality (1–3). During the 1918 “Spanish flu” pandemic, the vast majority (>95%) of fatal cases were complicated by bacterial pneumonias (4), with Streptococcus pneumoniae (Spn) accounting for the majority of bacterial infections. Data from more recent pandemics in 1957, 1968, and 2009 reveal a similar phenomenon; for example, in 2009 as many as 56% of patients tested positive for IAV associated bacterial pneumonias (5–8).

Given the significant disease associated with influenza virus and pneumococcus infections, considerable effort has been directed towards understanding the mechanisms responsible for bacterial outgrowth under these circumstances. These studies have revealed influenza-mediated alterations in the innate immune system that promote bacterial survival and growth, including decreased phagocytosis and loss of alveolar macrophages (1, 9, 10). Interestingly, in addition to increased bacterial burden, there is evidence that viral load in the lung is augmented following bacterial coinfection (11, 12), suggesting bacteria-mediated changes that promote virus infection and/or growth.

Pneumococcus in the lung is associated with a number of changes in the immune environment including the entry of neutrophils and macrophages as well as differentiation of T cells into Th17, Th2, and regulatory subsets, the last of which results in increased IL-10 (1, 13–20). In addition, bacterial products have the ability to directly modulate inflammatory responses. For example, pneumococcal components can reduce asthma associated inflammation by regulating effector function (21–23). Along with the immune modulatory effects on the lung environment that result from Spn infection, there is evidence that pneumococcus can directly impact T cell survival. For example, peripheral blood T cells from patients with bacteremia and sepsis exhibit high amounts of death (24–26). Further, in vitro studies show pneumolysin, the cholesterol dependent cytolysin produced by Spn, can induce T cell death (27). Based on these findings, we hypothesized that the entry and growth of Spn in the lung may impact the ongoing T cell response to influenza virus.

Clearance of acute influenza virus infection is dependent on the presence of a potent adaptive immune response. In support of this, severe cases of influenza infection in humans have been associated with the lack of an effective CD8⁺ T cell response in the lung (28). CD8⁺ T cells have been shown to mediate viral clearance through secretion of interferon γ (IFNγ) as well as cytolytic granule release (29, 30). A study performed during the 2009–2010 H1N1 pandemic found a strong negative correlation between the severity of symptoms and the number of IFNγ⁺IL-2⁻ CD8⁺ T cells (31), suggesting an important role for this cytokine in humans in the context of influenza.

Here we tested the hypothesis that Streptococcus pneumoniae negatively regulates the influenza specific CD8⁺ T cell response. We found a marked decrease in the overall size and quality of the influenza-specific CD8⁺ T cell response in the lung. The decrease in number was due, at least in part, to increased lymphocyte death following influenza virus infection. We also detected a decrease in the quality of influenza specific CD8⁺ T cells as evidenced by the reduced ability to co-produce IFNγ and TNFα in response to peptide stimulation. The decrease in cytokine producing cells was correlated with an increase in cells which exhibited cytolysis as their sole effector function. The selective inhibition of the production of cytokine was correlated with marked decreases in IFNγ mRNA. The altered influenza-specific T cell response appeared to contribute to disease in coinfected animals as reconstitution of the response by adoptive transfer significantly increased survival. The negative regulation of the influenza-specific T cell response observed in our study was not associated with changes in lung DC, but did correlate with an increase in Tregs. The changes in effector number and function were manifest predominantly in the lung, the primary site of bacterial and viral infection, suggesting high pathogen burden is necessary for the negative regulation of the influenza-specific CD8⁺ T cell response.

MATERIALS AND METHODS

Ethics Statement

All research performed on mice in this study complied with federal and institutional guidelines set forth by the Wake Forest University Animal Care and Use Committee. All studies were approved by the Wake Forest University Animal Care and Use Committee.

Bacterial and viral strains

S. pneumoniae EF3030 is a serotype 19F clinical isolate noted for its ability to colonize the nasopharynx for at least 21 days as well as its inability to cause invasive disease even when injected intravenously (32). S. pneumoniae was grown in Brain-Heart Infusion (Difco) broth supplemented with 10% heat-inactivated horse serum (Gibco) and 10% catalase (3 mg/ml) to an OD₆₀₀ of 0.8, correlating to approximately 1×10⁸ CFU/mL. Broth cultures were mixed 1:1 with a 50% glycerol solution and frozen at −80°C for future use. For CFU enumeration, S. pneumoniae was grown on tryptic soy agar (TSA) plates made with tryptic soy broth (Becton Dickinson) and 1.5% agar (Becton Dickinson) supplemented with 5% defibrinated sheep’s blood (Hemostat) and 4 μg/mL gentamicin (Sigma-Aldrich).

Influenza A/PR/8/34 (H1N1)

Virus stocks were grown and titered in fertilized chicken eggs (median egg infectious dose (EID₅₀)) essentially as described previously (33). Stocks were diluted in PBS, flash frozen, and stored at −80°C.

Mice

10–12 week old female BALB/c mice were purchased from The Jackson Laboratories. Mice were housed in a biosafety level 2 facility with access to food and water ad libitum.

Infection with PR8 and EF3030

Mice were anesthetized with Avertin (2,2,2-tribromoethanol) by intraperitoneal (i.p.) injection. Virus (10³ EID₅₀) or PBS as a control was administered via the intranasal (i.n.) route in 50μl of PBS. Four days postinfection, mice were anesthetized with Avertin and bacteria (10⁴ CFU) or BHI broth administered intranasally in 20μl. Disease was quantified using the following guidelines: 0 - No Disease; 1 - Ruffled Fur; 2 - Ruffled fur, limited mobility, slight hunching; 3 - Ruffled fur, ataxia, hunched, hypoxia, dehydration; 4 - Ruffled fur, ataxia, hunching, respiratory distress, hypoxia, dehydration. Animals receiving a disease score of 4 were removed from the study and humanely euthanized.

Quantification of EF3030 from lungs of infected animals

Aliquots of lung homogenates were serially diluted and plated on tryptose blood agar plates. Bacterial colonies were quantified following a 20 hour incubation at 37°C.

Quantification of viral burden

Viral RNA was extracted from lung homogenates using QIAamp Viral RNA Mini Kit (Qiagen). cDNA was synthesized from mRNA by reverse transcription using Superscript III RT kit (Invitrogen) and random primers (Invitrogen). For viral quantification, RNA primer-probe sets specific for H1N1 were used (BEI Resources). RT-PCR (qRT-PCR) was performed using the Applied Biosystems 7500 real-time PCR system.

Ex vivo Stimulation of T cells and analysis of effector function and death

At the designated days post influenza infection (d8 or d11), perfused lungs and mediastinal lymph nodes were isolated. Lungs were homogenized and incubated for 1hr at 37°C with collagenase D (100μg/ml) (Roche). Mononuclear cells were isolated by passage over a Histopaque gradient. MLN were mechanically disrupted and RBC removed by lysis with ACK. Cells were stimulated ex vivo with 10⁻⁷M NP_147-155 peptide in the presence of monensin and brefeldin A (BD Biosciences). Following a 5 hour incubation period, cells were stained for flow cytometric analysis. Fluorochrome-conjugated Ab detection reagents included: anti-mouse CD8α, LFA-1, and CD107a (all from Biolegend). APC conjugated NP_147-155/Kd tetramer was used to identify IAV specific CD8⁺ T cells (graciously supplied by the NIH tetramer facility). For cells stimulated with peptide, tetramer was included during the stimulation. This allowed tetramer labeling that otherwise may have been hampered as a result of TCR downregulation. Unstimulated samples were stained with tetramer during the surface stain only to circumvent tetramer induced cytokine production. Cells were then fixed and permeabilized (Cytofix/Cytoperm kit, BD Biosciences) followed by incubation with antibodies specific for IFNγ and in some cases TNFα. When 7-AAD (Biolegend) was used to determine cell viability, cells were incubated with 7-AAD following antibody staining. Cells were then washed extensively. For the subsequent detection of active caspase-3, cells were fixed and permeabilized (BD Biosciences) following 7-AAD staining. For the detection of T regulatory cells, the FoxP3 detection fix/perm kit (Biolegend) was used with PE conjugated α-FoxP3 antibody (Biolegend) together with CD4⁺ and CD25⁺ staining. Data was acquired using a FACS CantoII flow cytometer and analyzed using FacsDiva software.

Adoptive transfer

On d7 post influenza infection, mediastinal lymph nodes were harvested and stained with APC-conjugated Thy1-speciifc antibody (BD Biosciences). Thy1.2⁺ cells were isolated using anti-APC beads (Miltenyi Biotec) and MACS columns (Miltenyi Biotec) as per the manufacturer’s instructions. The purification was monitored by flow cytometric analysis. Population were 90–95% Thy1.2⁺. Coinfected animals that were 5.5d post influenza virus infection received either PBS or 2.6 x10⁶ isolated Thy1.2⁺ cells via tail vein injection. Animals were monitored through d11 for disease.

Quantitation of lung cell subsets

On d8 post virus infection, perfused lungs were homogenized and incubated for 1hr at 37°C with collagenase D (100μg/ml) (Roche). Single cell suspensions were passed over a Histopaque gradient to enrich for mononuclear cells. Recovered cells were stained with antibodies specific for CD11c, CD11b, B220, MHC Class II (all from Biolegend), and CD103 (BD Bioscience). Data was acquired using a FACS Canto II flow cytometer and analyzed using FacsDiva software.

Quantitation of cellular mRNA levels by real-time PCR

Isolated lymphocytes were stained with CD8 and APC conjugated NP_147-155/Kd tetramer. Tetramer⁺ and tetramer⁻ CD8⁺ lymphocytes isolated using a FACSAria cell sorter. Following a 5 hour peptide stimulation ex vivo, lymphocyte RNA was isolated by a standard Trizol (Invitrogen) phenol/chloroform extraction. cDNA was synthesized from mRNA by reverse transcription using Superscript III RT kit (Invitrogen) and random primers (Invitrogen). For IFNγ, perforin, granzyme B, IL-15, IL-18, TGFβ, and GAPDH mRNA analysis, commercially available Taqman primer-probe sets specific for the gene targets were used. RT-PCR (qRT-PCR) was performed using the Applied Biosystems 7500 real-time PCR system. Raw data values were normalized to GAPDH mRNA levels.

RESULTS

Coinfection with EF3030 and PR8 results in increased morbidity and mortality

To assess the potential for coinfection with Spn to regulate the ongoing anti-PR8 adaptive immune response, we developed a model that allowed survival of most animals to d11 postinfection with influenza virus. We used the well characterized mouse adapted influenza A virus A/Puerto Rico/8/34 [H1N1] (PR8) together with the pneumococcal strain EF3030. This strain is of the 19F serotype and is considered to lack the capacity to cause lethal invasive disease in mice (32, 34).

BALB/c mice received either PR8 (3.5×10³ EID₅₀) or PBS by the intranasal route (i.n.). Four days following PR8 infection, mice were given either 3.5×10⁴ CFU of EF3030 or brain/heart infusion media (mock) i.n. as a control (Fig. 1A). Animals were monitored daily for signs of disease. Animals receiving virus alone exhibited mild disease (disease score of 1–1.5) over the course of the infection (Fig 1B). In contrast, coinfected animals exhibited increased morbidity as early as d6 post PR8 infection (2 days post EF3030), with severe disease (score of 3 or greater) by d7 (Fig. 1B). Coinfected animals also exhibited significantly increased mortality, with roughly 50% succumbing to the infection by d11 post-PR8 infection (Fig. 1C). Animals receiving bacteria alone did not exhibit disease (Fig. 1B). While mortality was increased in the presence of coinfection, a large percentage of animals survive, thereby allowing assessment of the potential for Spn-mediated effects on the adaptive PR8-specific immune response.

Figure 1. Coinfection with EF3030+PR8 leads to enhanced disease and mortality.

A. Overview of experimental design. Animals infected as shown in A (singly infected with PR8, EF3030 or coinfected) were monitored for disease (B) and mortality (C). Disease was quantified using an adapted version of Tate’s disease scoring criteria which used the following parameters; coat ruffling, labored breathing, hypoxia, dehydration, ataxia and lethargy, conjunctivitis, and coat hygiene. Animals euthanized because of disease state were excluded from the disease score calculation after removal. Dotted line indicates day of bacterial infection. Data are derived from 26 (PR8), 44 (PR8+EF3030), or 20 (EF3030) animals assessed across 3 experiments. A Kruskal-Wallis non-parametric ANOVA with a Dunn’s posttest (B) and a Fischer’s Exact test (C) were used to determine significance. **** p<0.0001, *** p<0.0002

Coinfection with EF3030 and PR8 results in significant increases in viral and bacterial burden in the lungs of coinfected animals

One potential factor that could contribute to increased disease in coinfected animals was an increase in the level of bacteria and/or virus. To test this possibility, bacterial and viral load in the lung was determined on days 5, 8, and 11 post PR8 infection (d1, d4, and d7 post EF3030). Viral burden was assessed by real time RT-PCR following the protocol set forward by the Centers for Disease Control (35). Total RNA from the supernatants of lung homogenates was reverse transcribed using random primers to generate a cDNA library. Viral RNA was quantified using a primer probe set that targets hemagglutinin (HA) of H1N1 IAV. Using a DNA standard we calculated the viral RNA copy number as a measure of total viral burden. As shown in figure 2A, a significant increase in viral RNA was detected at 24 hours following delivery of the bacteria to PR8 infected animals compared to animals receiving PR8 alone. By days 8 and 11 post PR8 infection, virus was comparable in both groups.

Figure 2. Coinfection with PR8 and EF3030 leads to significantly increased viral RNA 24 hours after coinfection and significantly increased bacterial burden in the lung through d11.

Animals were infected as described in figure 1. Lungs were harvested for viral and bacterial load analysis at the indicated times post influenza infection. A. Viral RNA was extracted from lung homogenates using a QIAamp Viral RNA Mini Kit. cDNA was synthesized from mRNA by reverse transcription and subjected to qRT-PCR using primer probes specific for PR8 HA to enumerate viral RNA copies per lung. B. A portion of the lung homogenate was serially diluted and plated on tryptose blood agar plates for bacterial quantification. Data are the average of 24–29 individually analyzed animals/infection condition assayed over three independent experiments. A two-tailed student’s t test was used to determine significance. * p<0.05, ** p<0.01, *** p<0.001**** p<0.0001

As shown in figure 2B, on day 5 post PR8 (day 1 post EF3030) we observed an approximately three log increase in detectable bacteria in coinfected animals compared to animals receiving bacteria alone. Bacterial counts in coinfected animals remained high through d11 post PR8 infection. In contrast to coinfected animals, animals receiving EF3030 alone had begun to clear bacteria by d11, with a majority (5/9) of animals having bacterial counts below the limit of detection compared to coinfected animals where none (0/12) had cleared bacteria. Thus, coinfection resulted in increases in both bacterial and viral load in the lungs.

The number of influenza specific CD8⁺ IFNγ⁺ T cells in lungs is highly reduced in coinfected animals

CD8⁺ T cells play a critical role in the clearance of influenza virus. Given the high bacterial counts found in coinfected animals, we postulated that the presence of pneumococcus could alter the ongoing anti-influenza CD8⁺ T cell response. To assess the CD8⁺ T cell response elicited in these animals, we quantified the presence of IFNγ-producing CD8⁺ T cells specific for the influenza virus immunodominant epitope NP_147-155 in the lung and draining mediastinal lymph node (MLN). IFNγ was chosen because it is an important mediator of CD8⁺ T cell effector function in the context of influenza virus infection (30). By d8 post PR8 infection, a sizable virus-specific response was detected in the lungs of animals singly infected with PR8 (Fig. 3A). When lungs of animals coinfected with EF3030 were analyzed, a significant reduction (10.4-fold) in the total number of PR8-specific IFNγ⁺ CD8⁺ T cells in the lungs was detected (Fig. 3B). Assessment of the PR8-specific response in the MLN showed only a modest reduction (1.6-fold) in virus-specific cells in coinfected animals. The significant reduction in IFNγ⁺ cells in the lung remained at d11, although the decrease compared to singly infected animals was not as pronounced as at d8. The number of cells in the MLN was similar between the two groups at this time. These data suggest a strong negative regulation of the anti-influenza CD8⁺ T cell response at the site of active coinfection, i.e. the lung.

Figure 3. Coinfection with PR8 and EF3030 results in a substantial reduction in the number of IFN-producing NP-specific CD8⁺ T cells in the lungs and a modest reduction in the mediastinal lymph node.

Mice singly infected with influenza virus or coinfected were euthanized on days 8 and 11 post influenza virus infection. Perfused lungs and mediastinal lymph nodes were harvested. Isolated cells were stimulated ex vivo with influenza NP_147-155 peptide. Data shown are pre-gated on CD8⁺LFA-1^hi cells. A. Representative flow plots. B. Averaged data from 17–18 (d8) or 8–11 (d11) individually analyzed animals assessed over three independent experiments. A two-tailed student’s t test was used to determine significance. * p<0.05, **** p<0.0001

Increasing the number of influenza virus-specific T cells significantly increases survival of coinfected animals

We hypothesized the dramatic reduction in IFNγ-producing influenza specific T cells in the lungs contributed to the overall disease process during coinfection. To test this possibility, Thy 1.2⁺ cells were isolated from the MLN of influenza virus infected animals. Isolated cells were adoptively transferred into coinfected animals on d5.5 post influenza virus infection (1.5 d following Spn infection). This timeframe was chosen as it approximated the time at which effector cells can begin to enter the lung and critically is a timepoint at which mice remained healthy enough to undergo the transfer procedure. As above, mice were euthanized when they became moribund. As shown in figure 4, mice receiving influenza-specific T cells exhibited significantly increased survival. At d11 post influenza virus infection (5.5 days following the adoptive transfer), 41% of animals that received cells were alive in contrast to 8% of animals that received PBS. These data show that increasing the number of influenza virus specific T cells in coinfected animals promotes increased survival.

Figure 4. Adoptive transfer (AT) of influenza-specific T cells into coinfected animals results in significantly increased survival.

Thy1.2⁺ cells were isolated from the MLN of influenza infected on d7 post infection. Isolated cells were transferred into coinfected recipients on day 5.5 following influenza virus infection (day 1.5 following Spn infection). Animals were monitored and those that reached a disease score of 4 were euthanized. Results represent data from 2 independent experiments that together assessed a total of 12–13 mice per group. A Fischer’s Exact test was used to determine significance. * p<0.05

Coinfected animals exhibit a decrease in the number of NP-specific CD8⁺ T cells

The decrease in cytokine-producing cells could result from either a reduction in the absolute number of NP-specific CD8⁺ T cells or alternatively, from negative regulation of function in this population. To test the former possibility, lung and MLN cells from infected animals were stained with NP_147-155/Kd tetramer to quantify antigen-specific cells (Fig. 5A). While the percentage of tetramer⁺ cells was similar in the CD8⁺ population on d8 post PR8 infection, the total number of tetramer⁺ cells was significantly reduced (8.8-fold) in the lungs of coinfected animals compared to PR8 singly infected animals (Fig. 5B). Thus, the reduced number of IFNγ⁺ cells can be accounted for, in part, by a reduction in the number of NP-specific cells in the lung. The number of NP_147-155-specific CD8+ T cells in the MLN was not significantly different at this time although there was a decrease on average. Analysis of the response at d11 showed no difference in the number of NP-specific cells in either tissue (Fig. 5B). Given the observed reduction in the number of IFNγ-producing cells in the lung on day 11 post PR8 infection, there appears to be functional inactivation of lung cells at this time. These findings indicate a decrease in the number of NP-specific CD8⁺ T cells makes a significant contribution to the reduced IFNγ⁺ CD8⁺ T cell response observed in the lungs of coinfected animals.

Figure 5. Coinfection with EF3030 results in a substantial reduction in the number of NP-specific CD8⁺ T cells in the lung, but not the lung draining mediastinal lymph node.

Lungs and mediastinal lymph nodes from influenza virus infected or coinfected mice were isolated on day 8 and 11 post influenza and processed as previously described. A. Representative flow plots. Cells were pre-gated on CD8 and LFA-1^hi expression. B. Averaged data from 11–14 (d8) or 8–11 (d11) individually analyzed animals assessed over two independent experiments. A two-tailed student’s t test was used to determine significance. **** p<0.0001

Coinfection results in increased CD8⁺ T cell death in the lungs of coinfected animals

We hypothesized that the decrease in the overall size of the NP-specific CD8⁺ T cell population was due to increased death in the coinfected lung environment. To test this possibility, MLN and lungs were harvested from PR8 infected or coinfected mice at days 8 and 11 post PR8 infection. Isolated cells were stained for CD8 together with NP_147-155/Kd tetramer. We used a modified staining approach that allowed concurrent analysis of annexin V, active caspase-3 and 7-Aminoactinomycin D (7-AAD) to assess CD8⁺ T cell viability (36). By day 8 post PR8 infection, there was a significant increase in the percentage of tetramer⁺ CD8⁺ T cells in the lungs of coinfected animals that were positive for 7-AAD (Fig. 6A and B). There was no significant increase in 7-AAD⁺ cells in the MLN (Fig. 6A and B). Despite the decreased viability of the NP-specific CD8⁺ T cell population, increases in 7-AAD positivity did not correlate with the increased presence of active caspase-3 (Fig. 6C). This finding suggests death in coinfected animals is induced in a caspase-3 independent manner. These data support a model wherein the presence of EF3030 in the lungs of PR8 infected animals results in increased death of NP-specific CD8⁺ T cells.

Figure 6. Decreases in the NP-specific CD8⁺ T cell response in the lungs of coinfected animals on day 8 post influenza infection correlates with increased cell death.

Cells were isolated from the lung and MLN of animals infected with influenza virus or coinfected. Following staining for CD8 and tetramer, 7-AAD was added to identify cells that had lost membrane integrity as an indicator of cell death. A. Representative flow plots. Cells were pre-gated on CD8 and tetramer. B. Averaged data from 2–3 independent experiments assessing a total of 21–22 (d8) or 8–11 (d11) mice/infection condition. C. Cells were isolated from the lung and MLN of animals infected with influenza virus alone or coinfected. Following staining for CD8 and tetramer, cells were fixed, permeabilized and stained with antibodies specific for active caspase-3. The percentage of CD8⁺tetramer⁺ cells that stained positive for active caspase 3 is shown. Data are the average from 4–6 mice/condition. A two tailed student’s t test was used to determine significance. ** p<0.01, **** p<0.0001

The remaining NP-specific CD8⁺ T cell population in the lungs of coinfected animals exhibit an altered pattern of effector function

In addition to the production of IFNγ, effector T cells contribute to viral clearance through the production of TNFα as well as cytolysis. Thus, we determined whether the cells from coinfected mice that were incapable of producing IFNγ may retain the ability lyse or produce TNFα. Lymphocytes were isolated from the lung and MLN of singly or coinfected animals on days 8 and 11 following PR8 infection. CD8⁺ T cells were stimulated with peptide in the presence of tetramer to facilitate optimal detection by tetramer in subsequent analyses. IFNγ and TNFα production as well as CD107a at the cell surface, a surrogate for cytolysis, were assessed. This analysis revealed a significant decrease in the proportion of NP-specific cells in the lungs of coinfected animals that were polyfunctional (IFNγ⁺TNFα⁺CD107⁺) compared to cells isolated from singly infected animals (Fig. 7A). This decrease corresponded to a significant increase in the proportion of cells that exhibited cytolysis as their only effector function. The results in the MLN diverged from the lung, where there was no evidence of a shift towards cells that were only cytolytic (Fig. 7A). These data indicate selective regulation in the ability NP-specific CD8⁺ T cells from coinfected lungs to produce cytokine. By day 11, responses in singly and coinfected animals were similar in both tissues.

Figure 7. Coinfection results in qualitatively diminished CD8⁺ T cell function as measured by reduced polyfunctionality and co-production of IFNγ and TNFα.

Cells isolated from the lungs and MLN of animals infected with influenza virus or coinfected were stimulated with peptide as described previously. Anti-CD107a antibody was included in the stimulation phase to identify cells releasing lytic granules in response to peptide. Following the stimulation period, cells were stained for CD8, tetramer, LFA-1, IFNγ, and TNFα. Averaged data evaluating the distribution of the effector function of the influenza specific response is shown in A. The percent of NP-specific IFNγ-producing cells that co-produced TNFα is shown in B. Data are the average from 2–3 independent experiments assessing a total of 12–18 (d8) or 10–11 (d11) mice/infection condition. A two tailed student’s t test was used to determine significance. ** p<0.01, **** p<0.0001

We also determined the extent to which cells that were capable of producing IFNγ co-produced TNFα. We found a significant decrease in the percent of IFNγ producing cells in the lungs of coinfected animals that co-produced TNFα on d8 post PR8 infection (Fig. 7B). Thus TNFα production appeared to be most susceptible to the negative regulatory effects of EF3030 in the lung, followed by IFNγ. These data show that cytokine production is more susceptible to negative regulation by the presence of EF3030 compared to cytolytic function.

IFNγ mRNA is decreased in CD8⁺ NP-specific T cells isolated from the lungs of coinfected animals

To begin to understand the regulation of cytokine production at a mechanistic level, we determined whether the failure to produce cytokine was associated with a decrease in cytokine message. On d8 post PR8 infection, NP_147-155/Kd tetramer+ lung cells from singly or coinfected mice were purified by sorting. This approach allowed isolation of peptide-specific cells regardless of function. As such, the cells from coinfected animals represented a heterogeneous population of IFNγ producing and non-producing cells. Sorted cells were stimulated with NP peptide for 5 hours, RNA isolated and qRT-PCR performed with primer probe sets specific for IFNγ, perforin, granzyme B, and GAPDH. As shown in Figure 8, a significantly reduced level of IFNγ mRNA was detected in CD8⁺ NP⁺ cells isolated from coinfected compared to singly infected animals. In contrast, no decrease was observed in the level of mRNA for perforin or granzyme B, two of the primary components of cytolytic granules. These results mirrored the functional data that showed a population of cells from coinfected animals exhibit cytolytic potential in the absence of cytokine producing capability. These data suggest the lack of IFNγ production by NP-specific CD8⁺ T cells is regulated at a step prior to transcription.

Figure 8. The reduction in IFNγ production in NP-specific cells in the lungs of coinfected animals is correlated with decreased IFNγ mRNA.

On day 8 post influenza virus infection, CD8⁺tetramer⁺ and CD8⁺tetramer⁻ lung cells from singly or coinfected animals were isolated by FACS sorting. The sorted populations were cultured in the presence of NP_147-155 peptide for 5h to induce cytokine production. Following stimulation, mRNA was isolated and IFNγ, perforin, and granzyme B message quantified by qRT-PCR. Message levels were normalized to GAPDH. In all cases, the fold increase compared to the level of each mRNA present in tetramer negative CD8⁺ T cells from PR8 infected animals was calculated. Data are the average from two independent experiments assessing a total of 6 influenza or 7 coinfected mice. A two tailed student’s t test was used to determine significance. * p<0.05, ** p<0.01

Bypassing the T cell receptor by stimulation with phorbol 12-myristate 13-acetate (PMA) and Ionomycin (ION) does not induce IFNγ production in NP-specific CD8⁺ T cells from coinfected animals

CD8⁺ T cell responses are initiated upon engagement of the TCR and CD8 with cognate peptide antigen presented by MHC. TCR/coreceptor engagement induces signaling through src-family kinase pathways that eventually lead to transcription factors that drive cytokine production (37). One possibility to explain the lack of cytokine production in effectors from coinfected animals was alteration of the TCR signaling cascade such that cells could release granules, but not produce cytokine. We tested whether these cells were unable to appropriately initiate TCR signaling by assessing effector function following addition of phorbol 12-myristate 13-acetate (PMA) and Ionomycin (ION). These agents bypass the TCR by directly inducing PKC activation and calcium flux, respectively. Mice were singly or coinfected and on d8 post PR8 infection lung cells were stimulated with peptide or PMA/ION. As shown in figure 9, PMA/ION stimulation did not increase IFNγ production in the NP-specific cells isolated from the lungs of coinfected animals compared to that seen with peptide stimulation. These data exclude membrane proximal defects in TCR signaling as a mechanism to account for the failure to produce cytokine in cells from coinfected animals.

Figure 9. Stimulation with PMA/Ionomycin does not promote increased IFNγ production in influenza-specific CD8⁺ T cells isolated from the lungs of coinfected animals.

On d8 post virus infection, cells were isolated from the lungs of animals infected with influenza virus or coinfected and cultured in the presence of NP_147-155 peptide or PMA/ION. IFNγ production in the CD8⁺LFA-1^hi, tetramer⁺ population was determined. Data are the average from two independent experiments in which a total of 5 influenza or 8 coinfected mice were individually assessed. ** p<0.01

Differences in lung resident DC or IL-15 cannot account for the differences in lung effector cell function or survival in coinfected animals

Trans-presentation of IL-15 by lung dendritic cells has been shown to be an important signal for survival of effector cells (38). Thus, a possible contributor to the loss of cells in our model could be a reduction in lung DC capable of mediating this signal in coinfected animals. To determine whether this was the case, cells were isolated from the lungs of influenza virus or coinfected animals on day 8 following PR8 infection. DC and macrophage subsets in the lung were identified as follows- airway macrophages: high SSC CD11c⁺CD11b^lo/−, interstitial macrophages: high SSC CD11c⁺CD11b^int, recruited inflammatory macrophages: high SSC CD11c⁺CD11b^hi, airway DC: CD11c⁺CD11b⁻ CD103⁺Class II^hi, parenchymal DC: CD11c⁺CD11b⁺CD103⁻Class II^hi, monocyte derived respiratory DC: CD11c⁺CD11b⁺CD103⁻Class II^lo/int, plasmacytoid DC: CD11c^loB220⁺Class II^int. This strategy for subset identification is based on previously published results (9, 39). This analysis showed that there was no significant difference in the number of any of the DC subsets (Fig. 10A). Not surprisingly, we did observe a significant reduction in the number of airway macrophages in the lungs of coinfected animals (Fig. 10A), consistent with a previous report (9).

Figure 10. Differences in the number of lung dendritic cells or the cytokines IL-15, IL-18, and TGFβ cannot account for the negative regulation of effector cells in coinfected animals.

A. Lung dendritic cells and macrophages were quantified in virus infected or coinfected animals on day 8 post influenza virus infection. DC and macrophage subsets were identified as follows. Live cells were gated on CD11c. Macrophages were identified based on forward/side scatter profile and CD11b staining, with recruited inflammatory macrophages defined as high expressers of CD11b, interstitial macrophages as intermediate expressers, and alveolar macrophages as low to absent expressers. For DC subsets, plasmacytoid DC were identified by intermediate levels of MHC class II and B220 positivity. Airway DC were defined by CD103 and MHC Class II positivity. Inflammatory monocyte derived respiratory DC (MoRDC) were defined by low to intermediate expression of MHC class II and positive staining for CD11b. Parenchymal DC were defined as MHC Class II^hi/CD11b positive and CD103 negative. Data shown are the average of 9 influenza infected and 7 coinfected mice assessed across two experiments. B–D. Cellular RNA was extracted from lung homogenates on d8 following influenza virus infection. cDNA was synthesized from mRNA by reverse transcription and subjected to qRT-PCR using primer probes specific for IL-15, IL-18, TGFβ, or GAPDH. Fold changes for each animal were calculated based on comparison to the average level of each mRNA detected in PR8 infected animals. Data shown are the average of 15 coinfected and 15 influenza virus infected animals assayed across two independent experiments. A two tailed student’s t test was used to determine significance. * p<0.05

As noted above, the ability of lung DC to provide survival signals is dependent on trans-presentation of IL-15 (38). Thus it was possible that DC were present, but did not produce IL-15 for presentation to CD8⁺ effector cells. To address this, IL-15 mRNA was measured in the lungs of coinfected or virus infected animals on d8 post influenza virus infection. No decrease was observed in coinfected lungs. Instead, surprisingly, there was an approximately 2.5 fold increase in the amount of IL-15 mRNA in the lungs of coinfected animals (Fig. 10B). Together these findings suggest the increased death in effector cells from the lungs of coinfected animals is not the result of DC loss or the absence of IL-15-mediated survival signals.

Differences in IL-18 or TGFβ do not account for the differences in lung effector cell function or survival in coinfected animals

IL-18 has been shown to provide positive signals for CD8⁺ T cells cytokine production and survival (40, 41). In addition to provision of positive signals, cytokines can also inhibit function. TGFβ is one such well characterized inhibitory cytokine (42). Previous studies have shown that TGFβ is induced following infection with Spn (20, 43), making it an appealing candidate for negative regulation in the coinfected lung. IL-18 and TGFβ expression in the lung were assessed in singly and coinfected animals at d8 post-influenza virus infection. As shown in figure 10C and D, no difference in the level of mRNA for these cytokines was detected. These data suggest the loss of influenza-specific CD8⁺ T cells in our model is not due to the loss of the supportive cytokine IL-18 or an increase in the inhibitory cytokine TGFβ.

The lungs of coinfected animals have an increased number of Treg cells

Treg cells are another potent mediator of negative regulation. Previous studies have indicated that infection with Spn or exposure to pneumococcal components can increase Tregs in the lungs of mice (21, 23). To determine whether there was differential expansion/recruitment of Tregs in coinfected versus influenza virus infected animals, lung cells were analyzed for the presence of CD4⁺CD25⁺FoxP3⁺ cells (Fig. 11). In stark contrast to what was observed for CD8⁺ effector cells, a significant increase (4.2 fold) in the number of Tregs was observed in the lungs of coinfected animals.

Figure 11. Coinfected animals have a significantly increased number of FoxP3⁺ T regulatory cells.

Cells were isolated from the lungs of singly or coinfected mice on d8 post-influenza virus infection. T regulatory cells were detected based on positivity for CD4, CD25, and FoxP3. Data are the average of 9 influenza infected or 7 coinfected mice assessed across two independent experiments. A two tailed student’s t test was used to determine significance. * p<0.05

DISCUSSION

Infectious processes have the potential to significantly alter the lung environment. This is the product of signals resulting from tissue damage, cytokines/regulatory factors produced by tissue-resident cells, and innate immune cells that enter as a consequence of the presence of pathogen. Influenza virus is known to extensively modulate the lung environment. For example, infection results in the production of numerous inflammatory cytokines (including type I IFN, IL-6, MCP-1) (44), decreases in the ability of monocytes and neutrophils to phagocytose pathogens, and reduced mucosal ciliary action (1, 9, 10, 45).

Infection with pneumococcus also induces numerous changes to the lung environment. A robust neutrophilic infiltrate is one hallmark of Spn infection (46). Other immune modulatory signals present early after infection include the production of the pro-inflammatory cytokines TNFα, IFNα/β, IL-1β, and IL-6 (47). In addition, the presence of Spn components is associated with the activation of Tregs in the lungs (48). The complex array of immune signals present in this tissue as well as the direct action of bacterial components has the potential to regulate adaptive immune responses. While much attention has been focused on the influenza virus-mediated changes that promote increased bacterial outgrowth (9–12), to our knowledge the studies presented here are the first to address the impact of Streptococcus pneumoniae infection on the ongoing influenza-specific adaptive immune response. Investigation of this question led to the novel finding that the presence of pneumococcus negatively regulates the ongoing anti-viral CD8⁺ T cell response in the lung. Our data show that there is a marked decrease in the total number of CD8⁺ NP-specific T cells as well as a change in the distribution of effector function in these cells, with a pronounced shift away from cytokine producing cells to those that are exclusively cytolytic. Our findings suggest these changes in the effector population contribute to disease in coinfected animals as adoptive transfer of influenza-specific cells resulted in increased survival. These data would support previously published work demonstrating a strong correlation between the absolute number of IFNγ⁺CD8⁺ T cells and disease severity (31). Interestingly, the reduction in effector cell number was not associated with a prolonged period of increased viral load. Although higher levels of virus were detected in coinfected animals one day following Spn infection, by four days post Spn infection virus was similar in influenza virus infected and coinfected animals. Thus, there was not a direct relationship between virus load and disease. Previous studies have reported that under some circumstances damaging inflammation is not correlated with influenza virus load (49, 50). The increase in survival observed following adoptive transfer leads to the intriguing hypothesis that the reduced virus-specific effector cell number/function alters the balance of cytokines or inflammatory cells in the lungs of coinfected animals and this contributes to disease. Altered production of IFNγ by CD8⁺ effector cells has been shown to impact the inflammatory milieu present following influenza virus infection (51). In addition, TNFα has effects on multiple inflammatory cell types (52). While not yet evaluated, it is also possible that chemokine production is dysregulated in influenza-specific effector cells. CD8⁺ T cells can produce a number of chemokines that regulate the recruitment and function of a broad array of cells (53, 54). In disease processes where individuals survive past the initial phase of bacterial infection (24–48h), as is the case in our model, a reduction in cytokines/chemokines has the potential to significantly impact ongoing inflammation and/or the response to tissue damage. Additional studies are warranted to gain a fuller understanding of the role of influenza-specific effector cell regulation in the disease observed in our model.

The reduction in effector cell number in coinfected animals is in part the result of increased death in cells residing in the coinfected lung. The defect in cytokine production is correlated with a decrease in cytokine mRNA. The failure of PMA/ION to induce cytokine production suggests alteration of the membrane proximal TCR signal transduction pathway is not responsible for the lack of cytokine production. This finding would suggest these cells may be inherently incapable of producing cytokine. One possibility is that cells that are exclusively cytolytic preferentially survive in the coinfected lung. Given the death in our model, this is an attractive possibility that warrants further investigation. Alternatively, epigenetic changes induced as a consequence of signals present in the coinfected lung environment may have resulted in shut-off of cytokine production. Epigenetic regulation is a well described mechanism for the control of cytokine gene expression in T cells (55–57).

The question arises as to whether the negative regulation of the adaptive immune response is due to changes in the immune environment or a direct effect of bacterial components. Certainly immune signals in the form of cytokines have been shown to regulate the CD8⁺ T cell anti-viral response (38, 40, 41). For example, interleukin-15 production by lung dendritic cells has been identified as a critical signal for the survival of influenza-specific CD8⁺ effector cells in the lung (38). However, our findings would suggest that this is not the case in our model as IL-15 expression in the lungs of coinfected animals was higher than that in animals infected only with influenza virus. IL-18 has also been shown to promote generation and sustained presence of functional effector T cells. Cells with an ‘exhausted’ phenotype downregulate the IL-18 receptor, fail to produce cytokine, and have been implicated in the susceptibility to secondary bacterial infections (40). However, as was the case for IL-15, IL-18 expression was maintained in the lungs of coinfected animals and is thus also unlikely to be involved in the death of effectors or the decrease in cytokine production observed in our model. An opposing factor to the positive signals delivered by cytokines is the presence of negative regulators, i.e. TGFβ. This immunomodulatory cytokine is known to be increased following Spn infection (20, 43). Interestingly we did not observe any differences in the amount of detectable TGFβ in the coinfected versus influenza virus infected lung.

Finally we evaluated the presence of FoxP3⁺ Tregs. In spite of the decrease in influenza-specific effectors, there was a significant increase in Tregs in the lungs of coinfected mice. This was somewhat surprising given similar levels of TGFβ detected. One possibility is that cytokine production is impaired in these cells, similar to the reduction in cytokine in virus-specific effector cells. Alternatively, other sources of TGFβ may mask the contribution of Tregs when overall levels in the lung are assessed. Tregs employ a number of mechanisms to negatively regulate T cells, including production of adenosine, direct killing through the release of perforin and granzymes, and increases in intracellular cyclic AMP (for review see (58)). Whether these cells are directly involved in effector cells death or altered function will require further study.

In addition to the possibility for Tregs mediated regulation effectors, there is evidence that bacterial components can directly impact T cell function (22). In vitro studies suggest that the pneumococcal cholesterol dependent cytotoxin pneumolysin can lead to lymphocyte death in a Fas dependent manner when monocytes are present (27). Although this is an appealing model given the death observed in our system, fas mediated death would be expected to result in activation of caspase 3 (59), which was not detected in lung cells from coinfected animals. However, in the same report a necrotic death pathway could be triggered when monocytes were not present. This possibility is most consistent with the absence of active caspase 3 in dying cells observed in our model. The potential role of pneumolysin warrants further investigation. It is important to note however that despite sustained levels of bacteria at d11, the IAV-specific CD8⁺ T cell response is more comparable in both size and quality to that of animals singly infected with PR8. This could be explained in part by the fact that by d11 levels of virus are at their lowest. It is feasible that in our model that the increased death and alterations of CD8⁺ T cell effector function are dependent on signals from the presence of both pathogens. Alternatively, Spn is known to form biofilms as part of its infectious lifecycle (60). One consequence of biofilm formation is the reduction of autolysis that Spn uses to release virulence factors to the surrounding environment, such as pneumolysin (60). It is possible that EF3030 exists in a biofilm state by d11 as compared to d8 post IAV infection. If this were the case, it is possible that the bacteria would no longer be releasing components that could potentially modulate the IAV-specific CD8⁺ T cell response. Alternatively, this could indicate that bronchus associated lymphoid tissue (BALT) formation is impaired early by the presence of the bacteria. BALT has been shown to be a critical early component for the generation and maintenance of adaptive immune responses in the lung (61). If inducible BALT formation was inhibited by EF3030 this could lead to fewer IAV-specific CD8⁺ T cells in the lung at d8 post IAV infection. By d11 IAV-specific CD8⁺ T cells infiltrating the lung from the MLN could be reconstituting the IAV-specific CD8⁺ T cell response in the lung, which would be consistent with our data.

The consequences of Spn coinfection on the influenza-specific response may extend beyond the acute CD8⁺ T cell response studied here. For example, the reduced number of effectors may impact the number of tissue resident CD8⁺ memory T cells generated at the conclusion of the response. This could be exacerbated if the presence of Spn drove differentiation toward SLEC, which are unlikely to survive long term, further decreasing the memory pool. Spn-mediated increases in inflammatory cytokines, which are known to direct differentiation along the SLEC pathway (62), is an attractive mechanism for potential altered regulation of effector cell differentiation. Specifically, Spn infection subsequent to influenza has been shown to synergistically increase the level of type I IFN (63). In addition, it is possible that the regulation of the acute effector pool impacts the quality of the memory cells, e.g. their cytokine producing potential. A reduced or impaired tissue resident memory pool in individuals that experienced coinfection would increase their susceptibility to reinfection with IAV. This is currently an area of study. Finally, our finding of increased cell death in coinfected animals raises possibility that established tissue resident memory CD8⁺ T cells specific to other respiratory pathogens that are present in the lung during coinfection may be negatively impacted. Our results raise the interesting possibility that coinfection with IAV and pneumococcus could potentiate the deletion of already established cellular immunity in the lung environment.

In summary, our studies provide exciting new insights into polymicrobial disease states. Specifically, we show that coinfection with Streptococcus pneumoniae results in a marked decrease in the number of NP-specific CD8⁺ T cells in the lung. Further, surviving influenza-specific effectors exhibited altered effector function, i.e. the reduced production of cytokine, at the population level. Importantly, our data support a role for these changes in the high mortality observed following coinfection. The loss of IFNγ production in a subset of cells was associated with a decrease in the level of IFNγ mRNA. The negative effects on the influenza-specific effector population were independent of changes in lung DC populations and were not associated with decreased IL-15 or IL-18 or increased TGFβ expression. There was however, an association with increased T regulatory cells in the lungs of coinfected animals. These studies establish the ability of Spn to modulate the ongoing adaptive immune response to an existing pathogen. It is likely that the presence of coinfecting pathogens is common in the population and thus there is significant opportunity for cross-regulation of immune responses, similar to that reported here. Such cross-regulation is a potential contributor to pathogenesis and immunity in vivo.