Enhanced IL-1β production is mediated by a TLR2-MYD88-NLRP3 signaling axis during coinfection with influenza A virus and Streptococcus pneumoniae

Viral-bacterial coinfections, such as with influenza A virus and Streptococcus pneumoniae (S.p.), are known to cause severe pneumonia. It is well known that the host response has an important role in disease. Interleukin-1β (IL-1β) is an important immune signaling cytokine responsible for inflammation and has been previously shown to contribute to disease severity in numerous infections. Other studies in mice indicate that IL-1β levels are dramatically elevated during IAV-S.p. coinfection. However, the regulation of IL-1β during coinfection is unknown. Here, we report the NLRP3 inflammasome is the major inflammasome regulating IL-1β activation during coinfection. Furthermore, elevated IL-1β mRNA expression is due to enhanced TLR2-MYD88 signaling, which increases the amount of pro-IL-1β substrate for the inflammasome to process. Finally, NLRP3 and high IL-1β levels were associated with increased bacterial load in the brain. Our results show the NLRP3 inflammasome is not protective during IAV-S.p. coinfection.


Introduction
Secondary bacterial infections during influenza A virus (IAV) infection contribute to disease severity and mortality [1][2][3]. Streptococcus pneumoniae (S.p.) and Staphylococcus aureus (S.a.) are the dominant pathogens associated with IAV coinfection [1][2][3]. The coinfection of IAV and S.p. results in pneumonia due to multiple factors [4][5][6][7][8]. After IAV exposure, S.p. causes a severe infection requiring only a low inoculum size compared to a single infection [9]. IAV can also alter host immunological responses or lung homeostasis that can subsequently impede bacterial clearance [8,[10][11][12][13]. IAV infection enhances bacterial growth by depleting or PLOS  S. Animal Welfare Act of 1966. IACUC approval was obtained for the use of Ketamine and Xylazine for anesthesia. CO2 asphyxiation followed by cervical dislocation was the approved method for euthanasia. In addition to mice, embryonated chicken eggs (Charles River Labs) were infected with IAV at 10 days old for production of virus stocks.

Infectious agents
Mouse-adapted influenza A/PR/8/34 H1N1 virus (hereafter referred as PR8) stocks were propagated by allantoic inoculation of hen's eggs with seed virus. Plaque assays were performed using Madin-Darby canine kidney cells to confirm stock titer. Type 3 S.p. (ATCC 6303) was used in our studies. Colony Forming Units (CFU) assays were performed to confirm bacterial stock concentrations using brain heart infusion (BHI) agar.

In vivo infection
On day 0, mice were anesthetized by intraperitoneal injection with 80mg/kg Ketamine and 8mg/kg Xylazine diluted in PBS. Groups of 5-7 mice were infected with 125 PFU PR8 intranasally in a volume of 30μl PBS. Some of these groups were then mock infected and the others coinfected on day 7 with 1000 CFU S.p. intranasally in a volume of 30μl of PBS [9,51]. Additional groups of 5-7 mice were singly infected with 1000 CFU S.p. on day 7. At each time point, mice were monitor at least daily for weight loss and mice were euthanized by CO 2 asphyxiation followed by cervical dislocation when they achieved 30% weight loss or became moribund. Alternatively, mice were euthanized on day 9 or day 12 to collect lungs, liver, spleen and brain for examining lung pathology, cytokine levels by ELISA and flow cytometry, and for titering CFUs and PFUs. Viral titers from lungs that were homogenized by passing through a 70μm cell strainer were analyzed by plaque assay using MDCK cells as previously reported [52]. Quantification of S.p. from lung, liver, spleen and brain homogenates (also generated by passing through a 70μm cell strainer) was performed by making 10-fold serial dilutions of lung homogenate and plating 50μl on BHI agar plates and incubating in a 37˚C incubator with 5% CO 2 .

Histology
Lungs from coinfected mice collected on day 9 (2 d post-coinfection) or day 12 (5 d post-coinfection) were fixed in 10% neutral buffered formalin. Lungs were embedded in paraffin and 5μm sections stained with hematoxylin and eosin. Sections were examined and scored according to the scoring system in Table 1. Total lung pathology was the sum of all individual category scores for each animal. Histology slides were scored by Dr. Christopher Gilbert, a board certified pathologist at Cox Medical Hospital in Springfield, Missouri.

Immunoblotting
Lysates collected from in vitro infected BMDMs at different time points as described above (In vitro infection scheme and collection) were subjected to SDS-PAGE and gels were electrophoretically transferred onto polyvinylidine difluoride membranes (PVDF). Protein expression was examined using the following primary antibodies: anti-β-Actin and anti-IL-1β (D6A8, D3H1Z; Cell signaling technologies) were used with anti-rabbit HRP secondary antibody (Jackson Immuno Research, 111-035-144). Membranes were incubated in SuperSignal West Femto Maximum Sensitivity Substrate (ThermoScientific, 34096) and bands were visualized using an Azure Biosystems C300 imaging system.

Isolation of mRNA and real-time qPCR
Extraction of total mRNA was done using TRIZOL (Invitrogen). mRNA was then reversetranscribed into cDNA using a high capacity cDNA reverse transcription kit (Applied Biosystems, 4368814). cDNA samples were analyzed by real-time quantitative PCR (RT-qPCR) using DyNAmo HS SYBR Green qPCR Kits (Thermo Scientific, F410L) and relative values normalized to β-actin control. The following primer pairs were used: β-Actin

Statistical analysis
Student's t-test, one-way and two-way ANOVA with Tukey's or Dunn's post hoc analysis were performed using PRISM 6 from Graphpad. For weight loss during in vivo experiments, twoway ANOVA with Dunnett's post hoc analysis was performed using PRISM 6. In vivo survival analysis was performed using the Wilcoxon test using PRISM6. A p value <0.05 was considered statistically significant. Data are graphed as the mean +/-the SEM.
In addition, we examined pro-IL-1β expression to determine if increased signaling through PRRs during coinfection enhances priming signals. WT BMDMs were again infected with PR8 and S.p. alone or coinfected 3 hours apart. In samples collected 6 h, 12 h or 24 h after initial infection, we observed pro-IL-1β expression was enhanced during coinfection compared to singly infected samples (Fig 2B and S1 Fig). Pro-IL-1β expression was due to enhanced IL-1β mRNA (Fig 2C). In fact, mRNA levels of several cytokines were enhanced, including TNF-α, and IL-6 mRNA, in coinfected cells compared to single infected cells (Fig 2C). Overall, coinfection enhances transcriptional activation of cytokine genes.
We next examined the signaling pathways upstream that would regulate gene expression. During coinfection, the NOD2-RIPK2 pathway would respond to S.p. peptidoglycan fragment muramyl di-peptide (MDP), the RIG-I-MAVS pathway would respond to IAV uncapped RNA, and TLRs 2, 3, 7 and 9 would respond to their various ligands and activate TRIF or MYD88. Because all of these PRR pathways can facilitate NF-κB activation and cytokine gene expression through their adaptor proteins, we determined which pathways are involved in IL-1β production during coinfection by infecting BMDMs derived from WT, Ripk2 -/-, Trif -/-, Myd88 -/or Mavs -/mice. Intriguingly, Trif -/-BMDMs had higher IL-1β levels than WT BMDMs, suggesting TRIF signaling may play a regulatory role during coinfection ( Fig 2D). Importantly, only coinfected Myd88 -/-BMDMs had significantly reduced IL-1β compared to coinfected WT BMDMs (Fig 2D and 2E and S2 Fig). Finally, we found that Tlr2 -/-BMDMs had significantly impaired IL-1β production during coinfection compared to WT cells ( Fig  2F), demonstrating that a TLR-2-MYD88 signaling pathways primes pro-IL-1β during coinfection.

Pathways regulating IL-1β and the inflammasome in vivo during coinfection
Mice were infected with a non-lethal dose of 125 PFU of PR8 on day 0 and then mock infected or coinfected with a non-lethal dose of 1000 CFU S.p. on day 7. Another group of mice were singly infected with S.p. on day 7. Similar to infection in BMDMs, lungs from coinfected WT mice showed increased production of IL-1β, TNF-α, and IL-6 compared to PR8 or S.p. single infection of WT mice (Fig 3A-3C). Compared to WT coinfected mice, deficiency in either Nlrp3 -/or Myd88 -/had significantly decreased levels of IL-1β, and Myd88 -/mice also had significantly lower TNF-α levels (Fig 3A-3C). Although Myd88 -/mice lost more weight during single infection with PR8, there was no difference in mortality (Fig 3D and 3E), yet in the case of S.p. single infection, significant mortality was seen (Fig 3F-3G). During coinfection, Myd88 -/mice had higher weight loss and mortality than WT mice (Fig 3H-3I). Aim2 -/mice displayed a similar weight loss and mortality to WT mice (Fig 3H-3I). Finally, although Nlrp3 -/mice had similar mortality compared WT mice, their weight recovered earlier than any other genotype of mice (Fig 3H-3I).

Less bacteria in peripheral organs in Nlrp3 -/mice is associated with improved weight recovery
To understand the improved weight recovery seen in Nlrp3 -/mice and the increased susceptibility of Myd88 -/-, we examined viral and bacterial titers during coinfection. By day 9 (2 days post-coinfection), PR8 was cleared from the lungs of most mice, and there were no significant differences in viral titers (Fig 4A). S.p. titers were high on day 9 (2 days post-coinfection). However, there was a trend for Nlrp3 -/mice toward lower bacterial burden in the lungs, but  NLRP3 inflammasome during coinfection this did not reach statistical significance compared to WT mice (p = 0.0957) (Fig 4B). Examination of lung pathology on day 9 showed that Myd88 -/mice had slightly reduced overall lung pathology that approached significance (p = 0.0696) compared to WT mice (Fig 4C). This was mainly due to reduced cellular infiltrates into the lung (data not shown). Because differences in weight between WT, Myd88 -/and Nlrp3 -/mice did not occur until day 10 or later, we examined lung pathology and pathogen loads in mice on day 12 (5 d post-coinfection). There was improved overall lung pathology in Myd88 -/mice on day 12, particularly with respect to airway inflammation and lymphocyte numbers (Fig 4D-4G), but this was independent of lung bacterial numbers, as all genotypes of mice had similar lung bacterial loads on day 12 (Fig 4H). Bacterial numbers in the liver were also similar between all genotypes of mice (Fig 4I). Intriguingly, S.p. numbers in the brain were lower in both Nlrp3 -/and Myd88 -/mice compared to WT and Aim2 -/mice ( Fig 4J). However, Myd88 -/mice and Aim2 -/mice had more bacteria in the spleen than WT or Nlrp3 -/mice ( Fig 4K). Thus, earlier weight recovery in Nlrp3 -/mice was associated with lower bacterial burden in both brain and spleen compared to other genotypes of mice, but was independent of lung pathology.

Discussion
The invasion of bacteria like S.p. in IAV infected hosts is linked to increased death rates during pandemic outbreaks, such as the 1918 "Spanish Flu", where pneumococcus was found in samples collected from infected individuals [53][54][55][56]. Coinfections also occur during seasonal influenza epidemics to varying degrees [34,57]. Previous reports show that pro-inflammatory cytokines, such as TNF-α, IL-6, and type I IFN, increase during coinfection [21,22]. IL-6 and type I IFN display detrimental effects but TNF-α is protective during coinfection [21,22]. Thus, an improved understanding of the role for various cytokines and immune cells during coinfection is needed to understand this disease.
Some studies have examined IL1 receptor signaling during bacterial coinfection with IAV. Bansal et al. recently reported that Il1r1 -/mice are more susceptible to IAV-S.p. coinfection due to decreased alveolar macrophage numbers in Il1r1 -/mice [45]. However, Bansal et al. also found that Casp1 -/mice had similar survival to WT mice during IAV-S.p. coinfection [45]. This is in agreement with our findings where in Nlrp3 -/mice had similar survival to WT mice ( Fig 3I). However, these results suggest the inflammasome and IL-1β are not responsible for alveolar macrophage survival or mouse survival, or that IL-1β and IL-1α play redundant roles. Alternatively, Bansal et al. hypothesized an inflammasome independent mechanism for producing IL-1β [45]. A second group, Robinson et al., reported that Il1r1 -/mice are more susceptible to coinfection with IAV and a different bacteria, S.a., due to impaired Th17 responses [46]. However, there is a notable difference between our findings and those reported for Il1r1 -/mice infected with IAV-S.a. IAV-S.a. coinfection reduces IL-1β levels temporarily for the first 24 hours [46]. In our experiments, coinfection with IAV and S.p. only results in enhanced IL-1β levels. In a subsequent study, Robinson et al. treated IAV-S.a. infected mice with the NLRP3 inflammasome inhibitor MCC950 and found decreased S.a. numbers in the lungs, but similar survival compared to WT mice [58]. Similar to this second report by Robinson et al. [58], we found that IAV-S.p. coinfected Nlrp3 -/mice have lower bacterial numbers, but this was mainly in the brain and spleen, not the lung (Fig 4J and 4K).
Although these studies have contributed to our understanding of IL-1 signaling during coinfection, how the inflammasome is activated during IAV-S.p. coinfection was not well understood. Our findings demonstrate increased expression of pro-IL-1β during coinfection with IAV and S.p. is dependent on the TLR2-MYD88 pathway. Furthermore, we demonstrate that Nlrp3 -/mice or macrophages release less IL-1β than WT controls or Aim2 -/macrophages and mice. Because Nlrp3 deletion did not completely eliminate IL-1β production in vivo, other inflammasomes or pathways must be involved in IL-1β production in vivo during IAV-S.p. coinfection. One hypothesis is that a combination of NLRP3 and AIM2 contributes to inflammasome activation in vivo. Alternatively, other proteases in the lung, such as neutrophil elastase, may activate IL-1β [58]. Although Nlrp3 -/mice had only partially decreased IL-1β levels, we did observe improved weight recovery in these mice compared to WT mice. However, instead of reducing overt inflammation, as we originally hypothesized, Nlrp3 -/mice had similar inflammation to WT mice, in the lungs. Instead, on day 12, Nlrp3 -/mice had low bacterial burden in both brain and spleen, which was not observed in any other genotype of mice. This suggests the inflammasome and IL-1β may inhibit specific responses required for bacterial control, as eliminating NLRP3 improves bacterial burden. However, the mechanisms involved will require further investigation. As mentioned above, a previous report showed that mice treated with the NLRP3 inflammasome inhibitor MCC950 had decreased S.a. numbers during IAV coinfection, and this would support this hypothesis [58]. Interestingly, Myd88 -/mice in our experiments displayed decreased levels of IL-1β in the lung, decreased bacteria in the brain, and decreased lung pathology, yet they were more susceptible to coinfection. This would suggest that overt inflammation and tissue damage are not the only factors involved during coinfection [59].