Influenza-induced immune suppression to methicillin-resistant Staphylococcus aureus is mediated by TLR9

Bacterial lung infections, particularly with methicillin-resistant Staphylococcus aureus (MRSA), increase mortality following influenza infection, but the mechanisms remain unclear. Here we show that expression of TLR9, a microbial DNA sensor, is increased in murine lung macrophages, dendritic cells, CD8+ T cells and epithelial cells post-influenza infection. TLR9-/- mice did not show differences in handling influenza nor MRSA infection alone. However, TLR9-/- mice have improved survival and bacterial clearance in the lung post-influenza and MRSA dual infection, with no difference in viral load during dual infection. We demonstrate that TLR9 is upregulated on macrophages even when they are not themselves infected, suggesting that TLR9 upregulation is related to soluble mediators. We rule out a role for elevations in interferon-γ (IFNγ) in mediating the beneficial MRSA clearance in TLR9-/- mice. While macrophages from WT and TLR9-/- mice show similar phagocytosis and bacterial killing to MRSA alone, following influenza infection, there is a marked upregulation of scavenger receptor A and MRSA phagocytosis as well as inducible nitric oxide synthase (Inos) and improved bacterial killing that is specific to TLR9-deficient cells. Bone marrow transplant chimera experiments and in vitro experiments using TLR9 antagonists suggest TLR9 expression on non-hematopoietic cells, rather than the macrophages themselves, is important for regulating myeloid cell function. Interestingly, improved bacterial clearance post-dual infection was restricted to MRSA, as there was no difference in the clearance of Streptococcus pneumoniae. Taken together these data show a surprising inhibitory role for TLR9 signaling in mediating clearance of MRSA that manifests following influenza infection.


Introduction
Influenza viruses are single-stranded RNA viruses with a segmented genome capable of undergoing mutagenesis to evade host immunity and they cause seasonal outbreaks leading to over a half million deaths per year worldwide (World Health Organization, 2016) [1]. Influenza viruses can overcome traditional vaccine strategies such as inoculation with inactivated viruses, as this will not confer long-lasting protection to antigenic drift [2]. There are three types of influenza viruses that can infect humans (A, B, and C). Influenza A virus (IAV) and influenza B can both cause seasonal outbreaks but IAV is generally more severe. IAV infections can be complicated by bacterial pathogens including Staphylococcus aureus, and Streptococcus pneumoniae leading to increased morbidity and mortality [3]. Retrospective studies have shown that 95% of the deaths caused by the 1918 influenza pandemic (Spanish Flu) were complicated by bacterial superinfections [3,4]. More recently, whole-blood transcriptome analysis of over 225 influenza-infected patients showed a shift and enrichment in gene signatures from viral response to bacterial response in critically ill patients [5]. Thus, to decrease morbidity and mortality of IAV infections we need a better understanding of how to treat secondary bacterial infections. Past studies have found that IAV infections can lead to secondary bacterial infections by increasing the attachment sites for bacteria, reducing responsiveness of immune cells, and reducing efficiency of antibiotics [6][7][8]. Yet, the mechanisms underlying influenza-induced mortality are still poorly understood and we need better therapeutic strategies to improve outcomes for influenza-infected individuals.
Toll-like receptors (TLRs) are germline encoded pathogen recognition receptors (PRRs) capable of initiating innate immune responses, and regulating adaptive immunity to both viral and bacterial pathogens [9]. They are primarily expressed in immune cells, and are membrane bound and distributed in the extracellular membrane and endosomes making them able to recognize extracellular and intracellular pathogen components [10]. Manipulation of TLRs has shown great potential in combating bacterial infections. For example, agonistic stimulation of TLR4, a lipopolysaccharide (LPS) receptor, has been shown to improve bacterial clearance in Pseudomonas aeruginosa infected mice [11]. IAV has been shown to alter the expression of TLRs including downregulation of TLR2, a bacterial lipopeptide sensor, in human monocytes and dendritic cells [12]. TLR2 agonist stimulation was shown to have therapeutic potential to improve survival as well as bacterial and viral clearance in a mouse model of viral-bacterial coinfection [13]. However, little is known about the role of other TLRs that are altered in IAV infections and their implications in secondary bacterial infections.
IAV infection was shown to increase the expression of TLR9 in human monocytes and dendritic cells [12]. TLR9 is an endosomal receptor that recognizes unmethylated cytosine and guanine (CpG) motifs which are rich in viral and bacterial DNA and mitochondrial DNA (mtDNA) [14,15]. Here, we aimed to study the role of TLR9 in IAV-associated bacterial secondary infections, particularly with the bacterial pathogen methicillin-resistant Staphylococcus aureus (MRSA). Studying MRSA secondary infections is of high importance as in recent pandemics it was the main cause of secondary pneumonia in IAV infected individuals [16]. Additionally, MRSA is the leading cause of bacterial infections in humans worldwide, and infections are difficult to treat as MRSA is resistant to all known β-lactam antibiotics [17].
With the use of a mouse-adapted IAV strain, A/Puerto Rico/8/1934 (PR8), we found that TLR9 expression is elevated in lung macrophages, dendritic cells, CD8 T cells and epithelial cells from PR8 infected mice. TLR9 -/mice infected with PR8 or MRSA alone did not differ in clearance of either pathogen from wild-type (WT) mice, but they experience improved survival post PR8-MRSA dual infection and show improved bacterial phagocytosis and killing post dual infection Our findings show a previously unrecognized role for TLR9 in limiting clearance of MRSA post-dual infection.

IAV increases expression of TLR9 in macrophages
Changes in expression of different toll-like receptors (TLRs) (TLR2, TLR3, TLR4, TLR7,  TLR8, and TLR9) have been reported before in human monocytes and dendritic cells from seasonal influenza infected patients [12]. Similarly, in our murine experiments, we noted that TLR9 gene expression is increased in lung leukocytes obtained by collagenase digestion and ficoll density separation 5 days post-PR8 infection while TLR4 is reduced (Fig 1A). Protein expression was also increased in these lung leukocytes post-PR8 infection as measured by TLR9 immunoblotting (Fig 1B). To understand which cells were upregulating TLR9, we used flow cytometry to characterize the major immune cells in the lung compartment. We found that CD8 T cells, macrophages (interstitial and alveolar), and dendritic cells were the main cells with higher TLR9 expression post-PR8 ( Fig 1C). We did not find changes in NK cells, B cells, or neutrophils but CD4 T cells showed a lower frequency of TLR9+ cells (Fig 1C). Adherence selection of lung leukocytes for 1 h after collagenase digestion enriches for myeloid cells such as monocytes and macrophages and allowed us to detect gene expression changes in multiple TLRs after influenza infection in these cells. We found that apart from TLR9 being increased, TLR3 and TLR2 were also altered with increased and decreased expression, respectively ( Fig 1D). These later observations were also seen previously in human monocytes from influenza-infected individuals [12]. Interestingly, direct infection of isolated alveolar macrophages with PR8 also shows an increase in TLR9, TLR7 and TLR3, with no changes in TLR2, and reduced TLR4 (Fig 1E). We also detected TLR9 mRNA increased 24 hours post-PR8 infection in cultured bone marrow derived macrophages (BMDMs) (Fig 1F).
In order to determine whether TLR9 was being upregulated only in infected macrophages or also in non-infected cells, we infected bone marrow derived macrophages (BMDMs) for 24 or 48h with H1N1 or mock infection. We then measured levels of H1N1 infection by expression of the viral protein, NP, and looked for TLR9 levels by flow cytometry on cells which also expressed CD45 and F4/80 (Fig 2A). TLR9 is expressed on 55.4 ± 1.6% of H1N1 infected BMDMs compared to 50.2 ± 1.3% of mock infected cells at 24 h whereas NP expression was noted in only 0.3 ± 0.07% of cells at this time point (n = 4-5 group, P<0.05 for mock vs. infected TLR9%, flow plots from one sample for each shown). At 48 h, expression of TLR9 in mock-infected samples was seen in 19.9 ± 1.0% of mock infected cells and 28.2 ±1.08% of  (9,7,3,2,4) by RTqPCR and (B) western blotting of TLR9 and β-actin. RNA and protein were isolated from lung leukocytes post-collagenase digestion in mice infected with 100 PFUs of H1N1 (PR8) for 5 days or placebo (PBS). β-actin was used to normalize RNA in samples. (C) Frequency of TLR9 + cells measured by flow cytometry in lung immune cells post-collagenase digestion. Gating was as follows: CD4 + T cells (CD45 + ,CD90.2 + , CD3 + , CD4 + ), CD8 + T cells H1N1 infected cells. By 48 h, 0.1 ± 0.008% of BMDMs were NP+ (n = 4-5 per group, P<0.001 for TLR9% between mock and H1N1-infected cells at 48 h). Furthermore, BMDMs were not productively infected by H1N1 as NP expression decreased from 24 to 48h. To our knowledge, we are the first to report that infection of macrophages ex vivo with IAV can increase TLR9 expression. The signal to mediate this increase was likely independent of IAV recognition by TLR7, or by IAV-induced release of CpG rich mitochondrial (mt)DNA as TLR7 and TLR9 agonist stimulation lead to downregulation of TLR9 gene expression relative to mock infection ( Fig 2B). Taken together with the observation that TLR9 is increased on uninfected cells and on cell types not traditionally infected by IAV in vivo, these data are consistent with a secreted mediator being responsible for the upregulation of TLR9 expression post-H1N1 infection.

TLR9 -/mice show no difference in susceptibility to H1N1 or MRSA infection alone
The upregulation of TLR9 following H1N1 infection suggested the potential for cross-talk between viral infection and bacterial genome sensing. However, before addressing this, we wanted to first determine whether loss of TLR9 had any impact on host defense against infection with H1N1 alone or MRSA alone. Fig 3A demonstrates that 5 days post-infection with 100 PFU H1N1, WT (Balb/c) and TLR9 -/mice showed equivalent viral loads in the lung by plaque assay and in a separate experiment that viral M1 gene expression in the lung at this time was similar between genotypes ( Fig 3B).
Currently, there are conflicting results regarding the role of TLR9 in single MRSA infection [18,19]. To understand if TLR9 -/mice were susceptible to a single MRSA infection, we monitored mice for 7 days post-infection. We did not detect any deaths in either TLR9 -/or WT mice but weight recovery during the MRSA infection was slower in TLR9 -/mice (S1 Fig). We detected reduced immune cell infiltration in the alveolar compartment post-MRSA infection alone in TLR9 -/mice ( Fig 3C). However, this did not affect bacterial clearance ( Fig 3D) or lung injury ( Fig 3E) as both were reduced significantly 48 h post-infection in both genotypes. Previous work has shown that TLR9 -/mice have reduced TNF-α in the bronchoalveolar lavage fluid (BALF) post-MRSA [19]. We detected lower amounts of TNF-α, IL-6, IL-1β and IL-10 in the BALF of MRSA-infected TLR9 -/mice ( Fig 3F). Yet, lower amounts of these cytokines did not have a negative effect on bacterial clearance, lung injury, or survival. Reduced immune cell infiltration and the lower cytokine profile might be explained by lower NF-κB activation post-MRSA infection as TLR9 is a sensor of bacterial DNA [20].

TLR9 -/mice experience improved MRSA clearance post-dual infection
To determine whether the improved survival in the dual-infected TLR9 -/mice corresponded with better bacterial clearance, WT and TLR9 -/mice were infected with H1N1 on day 0 or were mock-infected. On day 5, mice received 7 x 10 7 CFU MRSA and lungs were harvested 24h later. Fig 5A shows that in WT mice, preceding H1N1 infection impairs clearance of MRSA relative to mice getting mock infection prior to MRSA. Furthermore, we were able to detect better bacterial clearance in the IAV-infected TLR9 -/mice 24 hours post-MRSA infection than in WT mice, but saw no difference in tissue injury (Fig 5B) or viral load (Fig 5C) in the presence or absence of MRSA between genotypes. Importantly, the difference we note in bacterial clearance at day 6 (24h post-MRSA infection) precedes the first deaths on day 7 ( Fig 4B).

TLR9 -/lung macrophages have increased phagocytosis, bacterial killing, and iNOS expression post-IAV infection
To determine if IAV was inducing changes in TLR9 -/mice that were leading to resistance to a MRSA infection, we measured the cytokine profile in the BALF in WT and TLR9 -/mice post-single IAV infection. There were no differences in cytokines tested 5 days post-IAV infection ( Fig  7A). We also found no significant differences in the immune cells recruited to the lungs (S4 Fig). Despite being present in equal numbers, monocyte/macrophages isolated from IAV-infected TLR9 -/mice have increased MRSA phagocytosis (Fig 7B). This increased phagocytosis correlates with higher expression of scavenger receptor A (SRA) on monocyte/macrophages isolated on day 5 from TLR9 -/versus WT mice infected with H1N1 ( Fig 7C). Our laboratory has previously shown that phagocytosis of non-opsonized MRSA requires SRA expression [23]. TLR9 -/mice also show improved intracellular killing of ingested bacteria ( Fig 7D) and TLR9 -/lung macrophages have higher levels of iNOS post-IAV compared to WT (Fig 7E). Nitric oxide production has been shown to be crucial in clearing MRSA infection [24]. To test whether TLR9 expression might suppress iNOS increase, we infected BMDMs from WT and TLR9 -/mice and measured iNOS expression. TLR9 -/macrophages have higher expression of iNOS post-IAV infection suggesting that TLR9 is a negative regulator of iNOS expression in BMDMs post-IAV infection (Fig 7F).

Loss of TLR9 in hematopoietic cells or antagonism of TLR9 in macrophages is insufficient to improve MRSA clearance post-H1N1
To determine whether loss of TLR9 just in hematopoietic cells was needed for the beneficial effects on MRSA clearance, we created chimeric (WT into WT and TLR9 -/in WT) mice and tested clearance of MRSA infection alone or clearance of MRSA following dual infection ( Fig  8A). Surprisingly, loss of TLR9 in hematopoietic cells alone showed no benefit in clearance of MRSA alone or MRSA post-H1N1. Further proof that inhibition of TLR9 just in monocyte/ macrophages was insufficient to improve MRSA clearance post-H1N1 is shown in Fig 8B where treatment of adherence purified monocytes and macrophages from H1N1-infeced mice with control oligodeoxynucleotide (ODN) or ODN2088, a TLR9 antagonist, demonstrated that ODN2088 treatment impaired (rather than improved) MRSA phagocytosis relative to control ODN-treated cells.

Lung epithelial cells upregulate TLR9 post-H1N1
Taken together, the results in Fig 8A and 8B suggest that stimulation of TLR9 on structural or other non-hematopoietic cells of the lung likely causes release of soluble mediators that  Cytokine measurement in BALF from BALB/c and TLR9 -/mice that were infected with IAV (100 PFUs, H1N1) for 5 days n = 5/group (some values are overlapping in the dot plots) for all but IL-10 which was from an experiment with n = 3 TLR9 -/mice and 5 Balb/c mice. (B) Ex vivo MRSA phagocytosis by macrophages isolated by collagenase digestion and adherence purification from mock or H1N1 infected mice on day 5, n = 3. (C) SRA expression analyzed by real-time RT-PCR in monocyte/macrophages isolated on day 5 from H1N1-infected mice, n = 3. (D) Ex vivo MRSA killing assay using adherence selected lung macrophages from BALB/c and TLR9 -/mice that were infected with IAV (100 PFUs H1N1) for 5 days or treated with PBS; n = 8/group in mock infections and n = 3/group in H1N1 infected mice. (E) Quantitative reverse transcriptase-PCR measurement of relative gene expression of iNOS from adherence selected lung macrophages from BALB/c and TLR9 -/mice that were infected with IAV for 5 days or treated with PBS; RNA samples were improve macrophage function in clearance of MRSA post-H1N1. To verify that TLR9 is modulated on lung structural cells post-H1N1 infection, we purified alveolar epithelial cells from WT or TLR9 -/mice and infected them ex vivo with MOI = 0.01 H1N1 or cells were mockinfected. After 24 h, RNA was made from infected cells and analyzed for TLR9 gene expression. Fig 8C demonstrates that epithelial cells from WT mice upregulate TLR9 mRNA expression and confirm that TLR9 -/mice do not express TLR9 transcripts. normalized to their β-actin levels and setting mock-infected Balb/c to 1; n = 3-5/group. (F) BMDMs from Balb/c or TLR9-/-mice were mock-infected for 24 h or were infected with MOI:0.01 H1N1 for 24 or 48 h before RNA was prepared and analyzed for expression of iNOS normalized to their β-actin levels and setting mock-infected Balb/c to 1; n = 3/group. Panels A and C were analyzed by student's ttest. Panels B, D, E and F were analyzed by two-way ANOVA with Sidak's multiple comparison test.; Non-significant (ns), � P<0.05, ��� P<0.001, ���� P<0.0001. Panels A-D are single experiments using multiple mice, panels E and F were repeated two times. https://doi.org/10.1371/journal.ppat.1007560.g007

Discussion
Lower respiratory infections are the fourth leading cause of death with 3 million deaths each year worldwide (WHO, 2018). The influenza virus infects the upper and lower respiratory tract and is successful at infecting 3-5 million individuals each year, taking the life of nearly a half million of these individuals [1]. Severe illness and death in influenza infections are seen mostly in high risk subjects, the very young and the elderly. However, recent influenza outbreaks have taken the life of younger and healthier citizens creating public health concerns. Secondary bacterial superinfections are responsible for high morbidity and mortality in influenza-infected patients [21]. Even with proper care including influenza vaccines, hygiene, and antibiotics, influenza-associated secondary bacterial infections are a burden to public health [25]. Additionally, the over-use of antibiotics has led to the selection of multidrug resistant bacterial pathogens making it harder to reduce the severity of bacterial infections [17]. Thus, we are in need of better therapeutic strategies against viral-bacterial co-infections that can improve public health. Influenza infections have been shown to alter the expression of TLRs in immune cells [12]. Manipulation of TLRs, in particular TLR2, has been shown to improve survival, and microbial clearance in mice co-infected with influenza and bacterial pathogens [13]. However, little is known about the potential roles that other TLRs can have in controlling viral-bacterial co-infections.
TLR9 is an intracellular receptor that recognizes unmethylated CpG motifs which are rich in microbial DNA [14]. TLR9 expression was reported to be elevated in monocytes and dendritic cells from influenza-infected patients compared to healthy individuals [12]. Here, we noted that IAV infection similarly increases TLR9 expression in murine immune cells from mice infected with a mouse-adapted IAV strain (PR8) (Fig 1A-1D). This increase can also be achieved in cultured alveolar macrophages ( Fig 1E) and BMDMs (Fig 1F) infected in vitro. We tested whether stimulation of TLR7, the innate influenza sensor could lead to increased expression of TLR9 as it is known that NF-κB activation by TLRs can induce TLR expression [26]. However, TLR7 stimulation actually reduced mRNA levels for TLR9 (Fig 2B). Influenza infections can lead to mitochondrial membrane permeabilization and release of mitochondrial components [27]. Release of mtDNA can also lead to activation of TLR9 due to mtDNA's high concentration of unmethylated CpG [15,20]. However, we found that CpG oligonucleotide stimulation of TLR9 also inhibited TLR9 mRNA (Fig 2B). Thus, it is still unclear how the IAV virus leads to the increased TLR9 expression noted in mice and humans. Because we see elevations of TLR9 in cells that are not actually infected with IAV, this mechanism is likely to be via secreted mediators and this will be a focus of our future investigations.
Mice lacking TLR9 (TLR9 -/mice) did not differ in viral response against IAV compared to WT as there was no difference in measured viral titers or M1 viral gene expression (Fig 3A and  3B), cytokine profiles (Fig 6A), or immune cell infiltration (S4 Fig). However, TLR9 -/mice were resistant to an IAV-MRSA coinfection with improved bacterial clearance (Fig 5A). The improved clearance of MRSA in TLR9 -/mice was not due to a preexistent resistance to the bacteria as there was no difference in single MRSA infection between WT and TLR9 -/mice ( Fig 3D). Previous reports focused on the role of TLR9 in single MRSA infection have shown conflicting results. TLR9 -/mice were reported to have decreased MRSA clearance despite showing a lower amount of TNF-α [19]. In contrast, MRSA was shown to induce a type I interferon response dependent on TLR9, and TLR9 -/mice were reported to have lower TNF-α and improved bacterial clearance [18]. Similar to the previous findings, we noted a decrease in cytokine secretion in TLR9 -/mice, specifically TNF-α, IL-6 and IL-10 were lower to single MRSA infection ( Fig 3F); however, TLR9 -/mice did not differ from WT in bacterial clearance, survival and tissue injury despite lower lung immune cell infiltration and cytokine release (Fig  3, S1 Fig). Thus, TLR9 seems to play a differential role in resistance to MRSA in the context of secondary bacterial infection post-IAV.
TLR9 -/mice have increased IFN-γ in the BALF together with higher numbers of IFN-γ producing cells (TH1, CD8 T cells, and NK cells) in the lung post coinfection (Fig 6A and 6B). This increase in IFN-γ provided a potential explanation for the improved bacterial clearance in TLR9 -/mice as IFN-γ has been shown to increase clearance of MRSA [22]. However, in our studies, INF-γ neutralization [confirmed by ELISA (Fig 6D)] was not able to decrease clearance of MRSA in TLR9 -/mice (Fig 6C). Therefore, the enhanced clearance of MRSA is independent of IFN-γ. Previous findings have shown that S. aureus is able to evade immunity and survive inside cells including phagocytic cells [17]. This is consistent with our data showing that IFN-γ cannot improve killing of MRSA post-H1N1 (Fig 6C).
While elevated IFN-γ was not critical for MRSA clearance, previous studies have shown that IFN-γ plays a negative role in Streptococcus pneumoniae (SPS3) clearance post-IAV infection by decreasing the expression of macrophage receptor with collagenous structure (MARCO) [28]. Just like MRSA, SPS3 is a gram-positive bacterial pathogen that is a high threat to influenza-infected individuals [29,30]. Thus, we wondered if TLR9 -/mice would make higher levels of IFN-γ following dual infection with H1N1 + SPS3, and if so, if that would correlate with higher SPS3 bacterial loads. Interestingly, we found that TLR9 -/mice have no difference in SPS3 clearance with or without an initial influenza infection (S5A Fig). Furthermore, IFN-γ levels in the BALF of TLR9 -/mice after IAV-SPS3 infection were not significantly higher than in dual-infected WT mice (S5B Fig). These findings highlight the important observation that TLR9 signaling has very different outcomes in the setting of influenza infection than in naïve mice and shows important distinctions in the mechanisms for susceptibility to MRSA vs. S. pneumoniae post-influenza.
Shortly after infection, MRSA is engulfed by phagocytes [31]. Macrophages, especially M1 (antimicrobial) polarized macrophages, play an essential role in the clearance of MRSA [32]. So we tested the ability of macrophages from TLR9 -/mice to clear MRSA in culture. TLR9 -/macrophages isolated from mock-infected mice had no difference in phagocytosis or intracellular bacterial clearance compared to WT (Fig 7B and 7D). However, macrophages from IAVinfected TLR9 -/mice are capable of improving bacterial clearance and killing (Fig 7B and 7D). The increased phagocytosis is likely related to the elevated SRA expression on macrophages from TLR9 -/mice post-H1N1 ( Fig 7C). Interestingly, TLR9 -/monocyte/macrophages from infected mice have higher expression of iNOS (Fig 7E). Similarly, BMDMs from TLR9 -/mice show higher iNOS expression following ex vivo infection with H1N1 (Fig 7F). It has previously been reported that mice lacking iNOS expression are deficient in clearance of MRSA and more than 50% of mice will not survive a MRSA infection past 24 hours [24]. Thus, induction of iNOS is a likely explanation for why TLR9 -/macrophages are more effective at clearing MRSA post-IAV infection.
Our results suggest that neutrophil accumulation is similar between WT and TLR9 -/mice in response to MRSA alone (Fig 3B) or following H1N1 infection (S4 Fig). In Fig 7 we show that lung monocyte/macrophages show improved phagocytosis and bacterial killing against MRSA in TLR9 -/mice. These results suggested that hematopoietic innate immune cells are primarily responsible for MRSA clearance. To determine whether the effects of TLR9 inhibition were localized to the myeloid immune cells, we created bone marrow chimeras to explore outcomes in mice which lacked TLR9 solely in the hematopoietic compartment. Interestingly however, these chimeras (WT into WT and TLR9 -/into WT) showed equivalent clearance of MRSA both alone and post-H1N1 (Fig 8A). This suggests that the beneficial effects of TLR loss may be due to non-hematopoietic cell signaling. In this regard, it is interesting that we have noted TLR9 upregulation on lung epithelial cells infected ex vivo with H1N1 (Fig 8C). We attempted to treat mice with 2088 ODN or control ODN to see if TLR9 antagonism in wildtype mice was beneficial. However, these results were variable at the highest dosage of ODN 2088 tested (50 μg given on days 0, 2 and 5 i.p.). We believe this reflects the fact that antagonism of structural or other non-hematopoietic cells is needed and thus our dosage may not have been optimal. Future experiments will explore WT into TLR9 -/chimeric mice for outcomes and will also explore lung-specific delivery of the ODN 2088.
In conclusion, our findings provide evidence that TLR9 plays a negative role in IAV-associated secondary MRSA infections. Blocking of TLR9 post-IAV infection can improve MRSA clearance and TLR9 -/monocytes/macrophages show increased bacterial phagocytosis and intracellular killing post-IAV infection. Taken together, this suggests that TLR9 antagonism may be an effective therapeutic for MRSA complicated influenza infections assuming proper dosing can be identified. However, care should be taken to know the nature of the secondary infection as TLR9 regulates MRSA, but not SPS3 coinfection. Future work will be focused on elucidating the mechanism(s) of influenza-induced upregulation of TLR9 and the pathways which TLR9 alters to regulate MRSA killing.

Mice
BALB/c mice were bred in the animal facilities at the University of Michigan (Ann Arbor, MI). Breeding colonies of TLR9 -/mice, on a BALB/c background, were kindly donated by Dr. Shizuo Akira [14] and were also bred in the animal facilities of the University of Michigan (Ann Arbor, MI). All mice used were at least 6-7 weeks old by the time of infection and/or treatment.

Chimeric BMT mice
Balb/c mice were treated with 9 Gy total body irradiation split dose and infused with 5 million whole bone marrow cells from either Balb/c or TLR9-/-mice. Chimeric mice were given acidified water (pH 3.3) for 3 weeks post-transplant and were used for experiments at 5 weeks post-BMT.

Bacteria and virus
Staphylococcus aureus (US300) was grown in nutrient broth and incubated with gentle agitation overnight at 37˚C. Streptococcus pneumoniae (SPS3) (serotype 3, 6303) was grown in Todd Hewitt Broth with 0.5% yeast extract and incubated overnight in anaerobic conditions at 37˚C and 5% CO 2 . Colony forming units (CFUs) were determined by optical density relative to known standard curves. Influenza A virus (IAV) (H1N1) strain A/PR/8/34 (PR8) was purchased from ATCC.

Cells
Alveolar Macrophages (AMs) were isolated by bronchoalveolar lavage performed with supplemented Dulbecco's Modified Eagle Medium (DMEM) (89% DMEM, 10% fetal bovine serum (FBS), and 1% penicillin-streptomycin (Pen-Strep) mixture) containing 5mM of ethylenediaminetetraacetic acid (EDTA). Total lung leukocytes were obtained by perfusing the lung with PBS followed by digesting the whole lung with Collagenase A and DNAse followed by Ficoll density separation as we have described [33]. Lung monocyte/macrophages were selected from these lung leukocytes by adherence to tissue culture plastic for 1 h and then cells were washed twice with PBS. Over 90% of attached cells were monocyte/macrophages (myeloid cells) by differential staining of attached cells, with the remaining cells largely neutrophils. Bone marrow derived macrophages (BMDMs) were obtained by differentiating bone marrow stem cells (from BALB/c or TLR9 -/mice) for 7 days with L-cell-supplemented culture medium (59% Iscove's modified Dulbecco's medium (IMDM), 30% L-929 cell supernatant, 10% FBS, and 1% Pen-Strep mixture). Primary alveolar epithelial cells were isolated using a procedure previously described [34]. Isolated epithelial cells were cultured on fibronectin coated plates for 2 days prior to infection with MOI = 0.01 PFU H1N1 for 24 h prior to harvest of cells for RNA. All cells were incubated in 37˚C in 5%CO 2 until used for experiments.

Model of infection
Influenza infections were done with intranasal instillation of 20μl of PBS containing 100 plaque forming units (PFUs) of PR8 to mice that were anesthetized with a mixture of ketamine and xylazine. Mock infections were PBS alone. MRSA infections were done intratracheally and the dose was always intended to be 7x10 7 CFUs per mouse but instillations through all the experiments came in the range of 5x10 7 CFUs-2x10 8 CFUs.

Survival assays
Mice were infected intranasally with 100 PFUs of PR8 for 5 days before intratracheal instillation of 7x10 7 CFUs of MRSA. All mice were anesthetized with a mixture of ketamine and xylazine before viral or bacterial infection. Mice were monitored daily during the course of infection and weighed each morning. Mice were euthanized when reaching a weight loss of greater than 25%.
In supplemental data, a survival assay was carried out to MRSA alone as well.

Plaque assay and viral M1 expression
Madin-Darby canine kidney (MDCK) cells obtained from Dr. Adam Lauring (University of Michigan) were used for the viral titer quantification from whole lungs of infected mice. Tetrazolium dye reduction assay of bacterial killing 2x10 5 alveolar macrophages (AMs) or BMDMs per well were seeded into duplicate 96 well plates: one control and one experimental plate. Cells from both plates were treated with IgGopsonized bacteria (MOI 50:1) for 30 minutes at 37˚C. The cells on the experimental plate were washed twice with PBS and then incubated with or without IFN-γ (10ng/ml) at 37˚C for 120 minutes, whereas the control plate was keep in 4˚C with 0.5% saponin in growth medium. After 120 minutes, 0.5% saponin in growth medium was added to experimental plate and Thiazolyl blue Tetrazolium Bromide assay was performed for each plate as explained in [35].
Opsonized phagocytosis values were obtained from control plates.

Non-opsonized phagocytosis assay
Heat inactivated and fluorescein isothiocyanate (FITC)-labeled bacteria was added into a halfarea black 96 well plate containing 2x10 5 cells per well at a MOI of 1:300 Cells were allowed to ingest bacteria for 120 minutes before trypan blue was added to quench extracellular fluorescence. Intracellular fluorescence was obtained by measuring fluorescence at 485ex/535em using a Spectra M3 microplate reader.

Enzyme-linked immunosorbent assay (ELISA) and albumin quantification
Cytokine measurement was performed with the use of R&D duo set ELISA kits for murine IFN-γ, TNF-α, IL-1β, IL-6, IL-10, and IL-17. Murine albumin measurement was performed with the use of the Bethyl Laboratory (Montgomery, TX) albumin ELISA kit.

Immunoblotting
TLR9 immunoblotting was performed in total lung immune cells after collagenase digestion.
In these experiments, cell lysates were obtained using RIPA buffer with protease inhibitor. Briefly, total protein from lung immune cells was separated in a polyacrylamide gel using a mini gel tank (Invitrogen, Carlsbad, CA), following by transferring protein to a polyvinylidene fluoride (PVDF) membrane that was blocked with 5% non-fat milk followed by overnight incubation with a polyclonal anti-TLR9 antibody(PA5-20202; Invitrogen; Carlsbad, CA).

Quantitative real-time PCR
mRNA was isolated using TRIzol according to the manufacturer's instructions. Relative gene expression measurements were achieved with the use of a Step-one plus real-time PCR system from Applied Biosystems (Foster City, CA). Gene-Specific primers and probes were designed with the GenScript Real-time PCR primer design software (Genscript Biotech Corporation, Piscataway, NJ). Table 1 shows the sequence of primers and probes used in the current studies.

Statistical analysis
Graphpad Prism version 7 software (Graphpad Prism Software Inc., La Jolla, CA) was used to analyze experimental results. When groups of two were compared, student's T-test was used to determine statistical significance. Groups of � 3 were compared using one-way analysis of variance with Bonferroni multiple mean comparisons.