In the United States, influenza results in 300,000 hospitalizations and 30,000 to 40,000 deaths each year (1
). While many people are infected with influenza virus annually, most recover without interventional therapy. However, in severe cases, bacterial superinfection is a common complication, causing increased morbidity and mortality. Tissue specimens from the 1918 influenza pandemic, which occurred before the advent of antibiotics, reveal a staggering 95% rate of bacterial superinfection (2
). That pandemic claimed an estimated 675,000 lives in the United States and millions of lives worldwide (1
). To this day, bacterial superinfection is responsible for over one-third of pediatric deaths from influenza each year in the United States (3
is considered the “classical” pathogen found during influenza superinfection; in postmortem cultures of lung tissue from victims of the 1918–1919 pandemic, 23.5% were found to have S. pneumoniae
, and another 8.1% were found to have Staphylococcus aureus
). The earliest murine models of influenza superinfection focused on the use of S. pneumoniae
, as it readily replicates in the mouse lung, and even a small inoculum will eventually kill mice (4
). However, many groups have been moving toward modeling superinfection with influenza and methicillin-resistant S. aureus
(MRSA), due to the recent rise in MRSA superinfections relative to S. pneumoniae
superinfections. In fact, from 2004 to 2012, including the 2009 H1N1 pandemic, 48% of bacterial isolates from pediatric patients who died from influenza were identified to be methicillin-resistant S. aureus
, with another 23% identified as methicillin-sensitive S. aureus
as well as 22% identified as S. pneumoniae
). Data for the current influenza season show that these trends continue. In the 2017–2018 season, one-half of pediatric patients who died from influenza tested positive for bacterial superinfection. A total of 54.6% of bacterial isolates from these patients were typed to be S. aureus
, with 50% of those isolates testing positive for methicillin resistance (6
). While mice clear MRSA more easily than S. pneumoniae
from the lung, immune responses to each bacterium differ in both humans and mice (7–9
Six to seven days after influenza infection, both mice and humans exhibit a “window of susceptibility” to bacterial infection (10
). The influenza virus infects respiratory epithelial cells, which in response produce high levels of type I interferon (IFN) (IFN-α and IFN-β) to induce an antiviral gene program in the lung (12
). Type I IFN has been shown to mediate susceptibility during influenza to superinfection with myriad bacteria, including the most common superinfecting pathogens, S. aureus
and S. pneumoniae
, as well as Gram-negative bacteria such as Escherichia coli
and Pseudomonas aeruginosa
). While type I IFN plays a significant role in the pathogenesis of bacterial superinfection, type III IFN (IFN-λ, interleukin-28 [IL-28], and IL-29) is produced at even higher levels than type I IFN in response to influenza (16
IFN-λ is thought to exert its antiviral effects through the same JAK/STAT pathway as type I IFN (17
) and similarly induces antiviral gene expression in the respiratory epithelium (19
). Researchers have recently proposed IFN-λ as a therapeutic for influenza (20
), demonstrating that mice treated early during influenza infection with pegylated IFN-λ2 (PEG-IFN-λ2) display reduced immunopathology and mortality during influenza infection. However, it was recently shown that mice lacking the receptor for IFN-λ display reduced bacterial burdens during influenza/S. aureus
). This observation, combined with the copious similarities between type I and III IFNs, suggests that IFN-λ therapy during influenza may exacerbate bacterial superinfection. Thus, we investigated the consequences of therapeutic IFN-λ on susceptibility to bacterial superinfection during influenza.
IFN-λ has recently been proposed as a potentially attractive therapy for influenza (20
). There is a dire need for more influenza therapeutics, as a broadly effective vaccine has not yet been developed, and current therapies are both time restricted and limited in effect (38
). With the continual mismatch of influenza vaccines to circulating viral strains (39
) and the ever-present threat of another influenza pandemic, more broadly effective treatments for influenza are certainly necessary. Galani et al. have shown that pegylated IFN-λ (PEG-IFN-λ) therapy during influenza reduces immunopathology and mortality by reducing the cytokine storm and neutrophil infiltration (21
). However, bacterial superinfection commonly complicates influenza, increasing morbidity and mortality. While reducing neutrophil recruitment to the lung ameliorates disease during influenza alone, these neutrophils are crucial for survival during Staphylococcus aureus
lung infection (40
). Moreover, mice lacking the IFN-λ receptor exhibit decreased bacterial burden during pulmonary MRSA infection as well as influenza/MRSA superinfection, suggesting that IFN-λ contributes to disease during bacterial infection of the lung (22
). Especially with the rise of drug-resistant pathogens such as MRSA as secondary bacterial pathogens, considering the risk for superinfection when evaluating a therapeutic for influenza is critical.
We report that overexpression of IFN-λ during influenza results in an increased lung bacterial burden upon superinfection with either MRSA or Streptococcus pneumoniae
. These data support findings by Planet et al. showing that mice lacking the IFN-λ receptor have lower bacterial burdens during pulmonary MRSA infection as well as influenza/MRSA superinfection (22
). Specifically, Planet et al. found increases in the expression of IL-22 and its associated antimicrobial peptide lipocalin 2 in IFN-λ receptor knockout mice. We have previously shown that exogenous lipocalin 2 decreases the bacterial burden during influenza/MRSA superinfection (23
). However, we found no change in IL-22 or lipocalin 2 expression upon IFN-λ overexpression. However, IFN-λ overexpression does not exactly recapitulate the phenotype of the total receptor knockout, as we see no difference in bacterial burdens during single S. aureus
infection (see Fig. S4 in the supplemental material), while in the IFNLR1 knockouts, the S. aureus
burden is decreased. It is also likely that there is a differential requirement for IL-22 and the associated antimicrobial peptide expression during transient IFN-λ overexpression versus constitutive IFN-λ receptor knockout. Thus, we investigated what else might be responsible for the acute IFN-λ-induced increase in the bacterial burden.
We demonstrate that IFN-λ decreases BALF neutrophil accumulation during influenza/MRSA superinfection. Galani et al. also showed a decrease in BALF neutrophils as well as total BALF cells, along with reduced peribronchial and parenchymal cell infiltration, upon PEG-IFN-λ administration during influenza infection alone. Notably, Galani et al. administered PEG-IFN-λ 2 days after viral infection, while we treated mice 5 days after viral infection, leading to a significant overexpression of IFN-λ at harvest 2 days later. While weight loss from our influenza virus infection mimicked the weight loss reported by Galani et al. with similar inocula (21
), treatment with IFN-λ 5 days after viral infection did not reproduce their reported reduction in viral burden. As current therapies for influenza decrease in effectiveness the later they are given during infection (38
), our delay in treatment until 5 days after viral infection is likely responsible for this discrepancy. We did not observe a decrease in BALF neutrophil accumulation either 24 or 48 h following influenza/S. pneumoniae
superinfection (Fig. S5), which suggests that the decrease in BALF neutrophils that we see in influenza/MRSA superinfection is not the cause of the increased bacterial burden.
Aside from differences in the timing of therapeutic intervention, bacterial superinfection drastically changes the immunological landscape of the lung. Interestingly, we saw no increase in bacterial burden upon IFN-λ treatment during MRSA pneumonia without preceding influenza (Fig. S4). This suggests that influenza-induced cytokines work in concert with IFN-λ during bacterial superinfection to reduce antibacterial immunity. Type I IFN is also broadly produced in response to influenza and strongly contributes to superinfection susceptibility (14
). We expect that the high levels of type I IFN in the influenza-infected mouse may synergize with our administered IFN-λ to produce this increase in bacterial burden that we see during superinfection but not bacterial infection alone. Many other cytokines are induced by influenza and play significant roles in its pathogenesis, including type 17 cytokines (23
) and IL-1 family members (42
). There is likely an interactive role for these players as well in the complex cytokine environment of the influenza-infected lung.
During bacterial superinfection, IFN-λ therapy produced no significant decrease in type I IFN expression or protein levels. We also saw no decrease in tumor necrosis factor alpha (TNF-α), IFN-γ, CCL3, or CCL4 levels, which were reported by Galani et al. to be reduced by PEG-IFN-λ administration. Instead, IFN-λ treatment during bacterial superinfection specifically increased neutrophil chemokines in the lung, while BALF neutrophils were decreased. This is consistent with previous findings that neutrophil depletion during S. aureus
pulmonary infection increases lung KC and G-CSF levels (23
), suggesting a “frustrated” chemokine production by the lung in response to a lack of neutrophils.
The reduction in BALF neutrophils implies that IFN-λ induces a defect in neutrophil production or migration. However, upon assessment of circulating blood cells, we saw no difference in neutrophils or total leukocytes. IFN-λ has been shown to suppress neutrophil migration in vitro
by transwell and EZTAXIscan assays (26
), suggesting a defect in the migration of neutrophils into airspaces where MRSA aggregates reside (43
). Surprisingly, while IFN-λ treatment decreased BALF neutrophils, lung neutrophils were not altered, as measured by flow cytometry. It must be noted that this flow cytometry was performed on lavaged lungs, which may explain the disparity between these two measurements. Importantly, bronchoalveolar lavage samples only the epithelial surface of the respiratory tract (44
), whereas flow cytometry was performed on a single-cell suspension digested from whole lung. Together, these data suggest that IFN-λ may specifically impair neutrophil migration across the lung epithelium.
Although IFN-λ treatment does not alter the number of neutrophils in the lung, it results in a marked decrease in neutrophil phagocytosis of MRSA in vivo. This effect of IFN-λ appears to be specific, as it alters phagocytosis but not myeloperoxidase activity or the expression of adhesion molecules. As the cytoskeletal changes of a cell responsible for in vivo migration mimic those necessary for phagocytosis, including actin remodeling and microtubule assembly, these data suggest that IFN-λ may be exerting a broader effect on neutrophil cellular motility and cytoskeletal rearrangement.
Together, these data strongly suggest that the use of IFN-λ as a therapeutic for influenza may result in adverse outcomes for patients who contract a secondary bacterial infection. We show that neutrophils, which are essential for the control of superinfecting pathogens (13
), are reduced in BALF following IFN-λ treatment. Moreover, neutrophil binding and uptake of MRSA are reduced with IFN-λ treatment, which correlates with an increase in the MRSA burden in the lung. Importantly, IFN-λ administration also decreases binding and uptake of S. pneumoniae
during influenza superinfection, which also correlates with an increase in the lung bacterial burden. Although IFN-λ may reduce influenza severity, both our data and the findings of other groups show that it worsens bacterial superinfection (22
). While new therapeutics targeting influenza are sorely needed, it is crucial that we take the risk of bacterial superinfection into account when evaluating new treatments.
MATERIALS AND METHODS
Six- to eight-week-old male C57BL/6 mice were purchased from Taconic Biosciences (Hudson, NY) and maintained under pathogen-free conditions. All studies were performed on age- and sex-matched mice. All animal studies were conducted with approval from the University of Pittsburgh Institutional Animal Care and Use Committee. All murine treatments (influenza virus, adenovirus, S. aureus, and S. pneumoniae) were administered by oropharyngeal aspiration.
Mice were infected with 25 PFU of influenza H1N1 A/PR/8/34 (41
) or phosphate-buffered saline (PBS). Five days later, mice were treated with 1 × 1010
viral particles (VP) of an adenoviral vector overexpressing mIFN-λ3/IL-28B (Vector BioLabs, Malvern, PA) or the enhanced GFP (eGFP) control in 50 μl of PBS (Genome Editing, Transgenic, and Virus Core, University of Pittsburgh). One day later, mice were challenged with 5 × 107
CFU methicillin-resistant Staphylococcus aureus
USA300 in 50 μl of PBS and harvested an additional 24 h later, or mice were challenged with 1 × 103
CFU Streptococcus pneumoniae
serotype 3 (ATCC 6303) in 50 μl of PBS and harvested 48 later.
Bacterial labeling and quantification.
FITC labeling of S. aureus was performed as follows. Bacteria were grown overnight to stationary phase with shaking at 37°C in casein hydrolysate yeast extract-containing modified medium (1806 CCY modified medium; ATCC). After measurement of the optical density at 660 nm (OD660), 10 μl of 10 mg/ml FITC in dimethylformamide (DMF) was added to 1 ml of bacteria and incubated with shaking at room temperature for 1 h. Following incubation, bacteria were spun at 10,000 × g for 5 min and washed with PBS twice. FITC labeling of S. pneumoniae was performed as follows. Bacteria were grown for 6 h with shaking at 37°C in Todd-Hewitt broth (BD Biosciences, Franklin Lakes, NJ). Next, 100 μl of this culture was used to inoculate a 100-ml flask of Todd-Hewitt broth and grown for an additional 12 h with continued shaking at 37°C. After measurement of the OD600, 10 μl of 10 mg/ml FITC in DMFO was added to 1 ml of bacteria, and the mixture was incubated with shaking at room temperature for 1 h. Following incubation, bacteria were spun at 10,000 × g for 5 min and washed with PBS twice. FITC-labeled bacteria were resuspended in PBS to bring the concentration to 5 × 107 bacteria per 50 μl. The bacterial burden was determined by plating serial 10-fold dilutions of the lung homogenate (right upper lobe of the lung homogenized in 1 ml sterile PBS).
Analysis of lung inflammation.
At harvest, mouse lungs were lavaged with 1 ml sterile PBS. This lavage fluid was centrifuged at 10,000 × g for 5 min to pellet cells, and the supernatant was frozen for cytokine measurement by an enzyme-linked immunosorbent assay (ELISA) for IFN-β and IFN-λ (mouse IFN-β or IFN-λ2/3 DuoSet; R&D Systems, Minneapolis, MN). Cell pellets from lavage fluid were resuspended in 500 ml sterile PBS and counted on a hemocytometer to enumerate infiltrating cells. These cells were then either processed by cytospin and stained for differential counting or centrifuged again at 10,000 × g for 5 min and then immediately frozen at −80°C for RNA extraction.
The right upper lobe of each lung was mechanically homogenized and plated for bacterial CFU counting, and cytokines in lung homogenates were analyzed with the Bio-Plex Pro mouse cytokine 23-plex array (Bio-Rad, Hercules, CA). The right middle and lower lobes of each lung were snap-frozen in liquid nitrogen for RNA extraction.
For assessment of myeloperoxidase (MPO) activity, the left lobe of each lung was perfused with PBS until white, snap-frozen in liquid nitrogen, and then mechanically homogenized for MPO activity assessment (MPO activity assay kit [colorimetric]; Abcam, Cambridge, UK).
To obtain a single-cell suspension, lungs were mechanically dissected and then incubated with shaking at 37°C for 30 min in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 1 mg/ml collagenase. After collagenase treatment, tissue was forced through a 70-μm filter and treated with ACK buffer to lyse erythrocytes. The resulting single-cell suspension was pretreated with anti-CD16/32 for 5 min to block Fc receptor binding before incubation with fixable viability dye and fluorochrome-conjugated anti-surface marker monoclonal antibodies (mAbs) for 30 min at 4°C. The following antibodies were used: anti-CD45, anti-CD11b, anti-CD11c, anti-Ly6G, anti-SiglecF, anti-F4/80, anti-CD24, anti-CD64, and anti-major histocompatibility complex class II (MHC-II). Live/Dead fixable aqua stain (Life Technologies, Carlsbad, CA) was used to determine cell viability. Samples were collected using an LSRFortessa flow cytometer (BD Biosciences, San Jose, CA) and analyzed using FlowJo software (vX.0.7; TreeStar, Ashland, OR). Flow gating began with doublet exclusion by comparing forward light-scatter area versus height and then debris exclusion by comparing forward versus side light scatter. Dead cells were excluded based on viability dye staining. Neutrophils were identified as CD45+ Ly6G+ cells.
Blood cell quantification.
Blood was taken from the heart via cardiac puncture upon tissue collection, placed into K2-EDTA tubes, and assayed within 30 min of recovery using a Hemavet 950FS hematology system (Drew Scientific, Miami Lakes, FL).
RNA was isolated from whole lung lobes snap-frozen in liquid nitrogen using the Absolutely RNA miniprep kit (Agilent Technologies, Santa Clara, CA), and its concentration was analyzed by spectrophotometry (NanoDrop ND-1000; Thermo Fisher Scientific) or isolated from −80°C frozen bronchoalveolar lavage cell pellets using the MagMax-96 total RNA isolation kit (Invitrogen, Carlsbad, CA). RNA was reverse transcribed into cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA), which was assayed by real-time PCR for gene expression with Assay on Demand TaqMan primer and probe sets (Life Technologies, Grand Island, NY).
Data were analyzed using GraphPad Prism 7 (GraphPad, La Jolla, CA). Analyses comparing two groups were performed by an unpaired t test with Welch’s correction, unless data were not parametrically distributed, in which case Mann-Whitney analysis was used. S. pneumoniae CFU data were log transformed before statistical analysis due to their distribution. Two-way analysis of variance (ANOVA) was used to compare repeated measures over time. Mortality data were analyzed by a log rank (Mantel-Cox) test. All figures show combined data from multiple replicate studies as means ± standard errors of the means (SEM). The indicated n values are numbers of animals per independent experiment. Statistical significance is indicated in the figure legends. P values of between 0.05 and 0.1 are displayed numerically.