Allergic inflammation hinders synergistic viral-bacterial co-infection in C57BL/6 mice

Asthma is a chronic airways disease that can be exacerbated during respiratory infections. Our previous findings that the inflammatory state of allergic airways at the time of influenza A virus (IAV) infection in combination with epidemiologic findings that asthmatics were less likely to suffer from severe influenza during the 2009 pandemic suggest that additional complications of influenza, such as increased susceptibility to bacterial superinfection, may be mitigated in the allergic host. To test this hypothesis, we developed a murine model of ‘triple-disease’ in which mice were first rendered allergic to Aspergillus fumigatus and co-infected with IAV and Streptococcus pneumoniae seven days apart. Significant alterations to known synergistic effects of co-infection were noted in the allergic mice including reduced morbidity and mortality, bacterial burden, maintenance of alveolar macrophages, and reduced lung inflammation and damage. The lung microbiome of allergic mice differed from that of non-allergic mice during co-infection. To investigate the impact of the microbiome on the pathogenesis of lung disease, we induced a perturbation with a short course of fluoroquinolone antibiotic that is often prescribed for lung infections. A significant change in the microbiome was complemented with alterations to the inflammatory profile and a drastic increase in pro-inflammatory cytokines in allergic mice which were now susceptible to severe disease from IAV and S. pneumoniae co-infection. Our data suggest that responses to co-infection in allergic hosts likely depends on the immune and microbiome states and that antibiotics should be used with caution in individuals with underlying chronic lung disease. Author Summary Asthma is a condition of the lungs that affects millions worldwide. Traditionally, respiratory infections are considered to have a negative impact on asthmatics. However, epidemiological data surrounding the 2009 influenza pandemic suggest that asthmatics may be better equipped to counter severe influenza including bacterial pneumonia. Herein, we introduce a novel mouse model system designed to recapitulate an influenza virus and Streptococcal co-infection in a host with fungal asthma. We found that underlying allergic asthma protects against severe disease induced by co-infection. Mice with underlying allergic inflammation had reduced damage to the lungs and did not show signs of respiratory distress. Among the differences noted in the allergic mice that were protected from viral and bacterial co-infection, was the lung microbiome. Allergic mice lost their protection from co-infection after we perturbed their lung microbiome with antibiotics suggesting that the lung microbiome plays a role in host immunity against invading pathogens.


49
Lung diseases are a leading cause of morbidity and mortality worldwide. Acute and chronic 50 respiratory diseases, excluding infections, affect greater than 12% of the population in the United States 51 and hundreds of millions worldwide (1). Asthma is the most prevalent of these (2) and has the greatest 52 economic burden (3), in addition to being one of most challenging lung conditions to investigate and treat. 53 While the exact etiology of asthma remains unstipulated, specified phenotypes based on symptoms, 54 immunologic profiles, genes, and environment, are confounded by gender and age. Furthermore, asthma 55 exacerbations can be triggered by respiratory viral infections (4, 5) and some reports suggest that 56 asthmatics are at risk for bacterial pneumonia (6, 7).

57
Over four million deaths every year (predominantly children <5 years) are caused by acute 58 respiratory infections (8). Influenza and Streptococcal pneumonia contribute to approximately one million 59 hospitalizations annually in the U.S. (9, 10). While influenza alone can be fatal, recovering patients have 60 an increased susceptibility to respiratory bacterial infections (11), of which Streptococcus pneumoniae 61 (Spn), a pathogen associated with community-acquired pneumonia (12), is highly associated with severe 62 disease during influenza infections. The 'Spanish Flu' pandemic exemplified this predilection with the 63 majority of deaths attributed to subsequent bacterial infections (13), as did the 2009 Swine Flu pandemic in which 29-55% of deaths resulted from secondary bacterial pneumonia (14,15). The occurrence of 65 pulmonary infections in asthmatics is augmented by high disease incidences of each, and seasonal overlap 66 between infectious agents and allergens. 67 Although asthma was a risk factor for hospitalization during the 2009 'Swine Flu' pandemic (16). 68 subsequent studies noted that some asthmatics had less severe outcome (including reduced bacterial 69 pneumonia) compared to non-asthmatics (17)(18)(19)(20)(21). Explanations for this unexpected and counterintuitive  Allergic airways inflammation protected mice against severe disease from co-infection. 103 Mouse model systems that can simulate complex interactions between allergy and respiratory 104 infections are limited, but important to study disease-disease interactions that may alter host responses.

105
Since respiratory infections with viruses and bacteria are considered triggers for the development of 106 asthma, the previously reported animal models utilized infectious agents prior to allergen provocation 107 (30). While asthma can indeed be triggered by respiratory infections, it can also be exacerbated by the 108 same (31, 32). Herein, our first goal was to develop and characterize a model system in which 109 respiratory infections occurred in established allergic airways disease. Mice were subjected to A. 110 fumigatus allergen sensitization and challenge (25,33), and then infected with IAV (23), followed by S. 111 pneumoniae (Fig 1A). Naïve mice were used to measure baseline, while asthma-only, influenza-only 112 (Flu Ctr), bacteria-only (Bact Ctr) mice served as single disease controls. Dual condition groups included Asthma+Flu (AF), Asthma+Bact (AB), and Flu+Bact (FB), while Asthma+Flu+Bact (AFB) 114 triple-disease condition served as the experimental group.

115
As previously demonstrated by us (23), IAV infection during peak airways inflammation did not 116 induce weight loss in allergic mice (AF group, Fig 1B), whereas the same dose of virus triggered about a 117 12% weight loss in non-allergic mice (F group, Fig 1B). Non-allergic mice that were co-infected (FB 118 group) lost ~20% weight at the termination point in this study (Fig 1B), and continued to lose weight 119 and succumbed to disease by 6 dpi with Spn (data not shown). In stark contrast, allergic mice that were 120 subsequently co-infected (AFB group) did not lose weight and had a comparable weight profile to the 121 AF group (Fig 1B) and 83% in the AFB group survived (data not shown). As such, although mortality 122 was not an output of this study, allergy appeared to delay/protect mice from IAV+Spn-induced mortality 123 which may provide a time advantage for clinical therapeutics. Infectious virus was absent in the lungs of 124 mice in all groups at 3 dpi (data not shown) which differs from previous studies that have demonstrated 125 a viral rebound after Spn co-infection, albeit using the laboratory strain of IAV (34). The bacterial 126 burden in the allergic lungs was not sufficient to visualize by fluorescence like in co-infection alone (Fig   127   1C), but conventional enumeration of pneumococci on blood agar showed significant reduced bacterial 128 loads in allergic mice compared to the non-allergic co-infected mice (Fig 1D). Bacterial dissemination 129 into the blood may also be delayed/reduced in allergic mice (Fig 1D).

131
Allergic mice had a more diverse immune cell signature in the airways although tissue inflammation 132 was lower compared to non-allergic mice during co-infection 133 Inflammation is an important hallmark of both allergic disease and respiratory infections, 134 although the cell types that dominate are different. We measured the number and type of leukocytes in 135 the airways (bronchoalveolar lavage, BAL) as a marker of disease severity. As expected, inflammatory 136 cells were increased significantly over steady state (3.70×10 4 ± 1.04×10 4 live cells) after each trigger 137 (Fig 2A). Macrophages were more abundant in single infections with and without allergy, but significantly lower in the FB group (Fig 2B) as previously demonstrated (35). Eosinophils, B cells, and 139 CD4 + T cells followed a similar pattern of abundance, wherein cell numbers were significantly higher in 140 pathogen infected allergic mice than their non-allergic counterparts (Fig 2B). Neutrophil infiltration was 141 markedly higher in the FB group, while CD8 + T cell recruitment was similar except for the asthma-and 142 bacteria-only controls which had very low influx into airways (Fig 2B). 143 Over exuberant immune responses are considered a mechanism by which synergistic actions of 144 IAV and Spn increased host morbidity and mortality (36). Analysis of hematoxylin and eosin stained 145 sections of lung tissue showed that widespread parenchymal inflammation was present in IAV-infected 146 mice (Flu, Fig 3) but that Spn-infected mice had minimal areas of inflammation (Bacteria, Fig 3) most 147 likely due to effective bacterial clearance in otherwise healthy hosts. In contrast, extensive areas of lung 148 parenchyma in FB group mice were consolidated by fluid exudates and inflammatory cells consisting 149 mostly of neutrophils and macrophages (Fig 3). As expected, inflammation in the asthma-only control 150 mice mostly surrounded the terminal airways (Fig 3), and similar inflammatory foci were observed 151 around the small airways in both the AF (Fig 3) and AB groups (Fig 3). Significantly, pulmonary 152 lesions in the AFB group mice (Fig 3) were much less severe than those in the FB group (Fig 3), and 153 resembled those of AF mice (Fig 3). Histopathologic scoring of diffuse alveolar damage markers such as 154 alveolar inflammation and protein/fibrin deposition were all much higher in the FB group than in the 155 AFB. The higher levels of mucus production in all allergic mice, irrespective of the presence or type of 156 infectious agent, correlated with the reduced damage and loss of bronchiolar epithelium in these lungs.  Mucosal microbial abundance in the allergic co-infected mice differed from baseline 161 The importance of the microbiome in health and disease is increasingly recognized (37). Our 162 understanding of the contribution of endogenous microbes in the gastrointestinal system is more advanced than that of the respiratory system, partially due to the comparatively low biomass and 164 sampling difficulties. Therefore, studies that incorporate the microbiome into disease pathogenesis are 165 important. Herein, we investigated the microbial diversity in the BAL and lungs by comparing the 166 microbiome abundance in each niche of each group in comparison to naïve controls at the genus level 167 (Table S1). Differential abundance changes of microbiota under different conditions although the 168 changes were not significant (p>0.05) after multiple test correction across all comparisons. Micrococcaceae were reduced in BAL of mice infected with Spn (p<0.05). Skermanella dominated in 182 the lungs of Bacteria-only control mice while Rothia and Fusobacteria were reduced (p<0.01) (Fig 4C). 183 Clear separation of clusters were identified between AF and naïve mouse BAL samples wherein 184 Ruminococcus, Allobaculum, S24_7 of Bacteroidales and Clostridiales were enriched in AF (p-value < 185 0.01) and multiple taxa were depleted especially Ruminococcus (p=0.005). (Fig 4D). Minor overlap was 186 observed between the lung microbiota of AF and naïve mice with Skermanella, Ruminococcaceae of Clostridiales, S24_7 of Bacteroidales and Nitrospira highly enriched while Propionibacterium and 188 Corynebacterium were reduced (p=0.01) in AF group (Fig. 4D). 189 Microbial clusters in the BAL and lungs of AB mice were slightly separated from naïve with 190 enrichment of Oscillospira of Ruminococcaceae (p<0.01) and depletion of Nitrospira (p<0.01) in the 191 AB group (Fig 4E). Tight clusters showed clear separation of lung microbiota between the FB group 192 and naïve mice where Streptococcus was highly enriched in the BAL (p<0.01) and >20 taxa were 193 depleted in both BAL and lung (p<0.05) in the FB group (Fig 4F). Similarly, clear separation of clusters 194 between the AFB group and naïve animals was observed in both niches (Fig 4G). Bacillaceae,195 Streptococcus, Turicibacter, and Bacteroides were enriched in BAL of AFB (p<0.05) while Antibiotic (Abx) overuse is a growing concern with both short-and long-term implications. We 202 hypothesized that a microbiome dysbiosis induced by Abx treatment will increase synergistic 203 pathogenesis of IAV and Spn in allergic hosts. Mice were treated daily for two weeks with levofloxacin, 204 a commonly used Abx for respiratory infections, prior to virus infection (Fig 5A). Antibiotic-treated 205 IAV-infected mice (Flu ctr) had significantly lower nadir than untreated counterparts (Fig 5B). Abx 206 treatment did not alter weight curves in the other groups except the triple-disease state (AFB) in which 207 case allergic co-infected mice treated with antibiotics exhibited weight loss similar to Flu control mice 208 (Fig 5B). A concomitant increase in the bacterial burden in the AFB group occurred in all three niches 209 tested while the increase in the FB group did not reach statistical significance in the blood (Fig 5C). 210 While the total cell numbers were reduced in all groups after levofloxacin treatment (Fig 6A) 211 compared to untreated mice (as shown in Fig 2), the cell types most dramatically altered were macrophages and CD8 + T cells (Fig 6B). Most cell types were drastically reduced after Abx treatment, 213 while neutrophils were increased in the Flu control and FB groups while CD4 + T cells and CD8 + T cells 214 were increased in the Flu control group (Fig 6B). Interestingly, eosinophils were reduced in all groups, 215 but most notably in the AB and AFB groups (Fig 6B). When we normalized the cell types that were 216 measured in each group, the immune profiles were notably of different compositions between groups 217 (Fig 6C) and in comparison untreated mice. 218 We investigated the pro-inflammatory cytokine milieu in the BAL and the lungs to better  Table 1). Similar trends were observed in the lung homogenates ( Table 1). Treatment of mice with 223 Abx affected the cytokine profile in both the BAL and lungs with significant increases occurring in both 224 the FB and AFB groups (Table 1). However, even after Abx treatment, most cytokines were 225 significantly more highly expressed in the FB compared to the AFB group (Table 1).

227
Mucosal microbiome diversity was reduced by antibiotic treatment. 228 In order to determine if we were successful in inducing dysbiosis, we analysed the microbiome 229 in the BAL and lungs of mice that were treated with Abx in comparison to untreated mice. The PCoA 230 ordination indicated that microbiota of Abx-treated mice formed independent clusters from untreated 231 mice except in the AFB group (Fig 7). Differential abundance analyses indicated that the majority of 232 treated mice had reduced taxa abundance. While Facklamia, Bacillaceae, and Enterococcus (p=0.046) 233 were enriched in BAL of Abx-treated Asthma group, significant depletions in were identified in multiple 234 genera especially Alphaproteobacteria and Actinobacteria (Table S2). Distinct clusters in the lungs of 235 asthma group (Fig 7A) showed significant enrichment for Anaerococcus of Clostridia and Lactobacillus in the Abx-treated group, while numerous genera including Proteobacteria, Firmicutes and 237 Actinobacteria, had decreased abundance after Abx treatment.

238
A majority of taxa in the microbiome of Abx-treated Flu control mice were reduced except for a 239 few genera (Peptoniphilus, Selenomonas, and Enterococcus) that were enriched in the BAL clusters (Fig   240   7B). Fusobacterium enriched the lung microbiome cluster of Abx-treated Flu controls while 241 Streptococcus enriched both BAL and lung. All identified taxa in the microbiota of Bacteria-only 242 controls had a significantly reduced abundance after Abx-treatment except for Streptococcus that was 243 enriched in the lungs with the clearly separated clusters (Fig 7C). 244 Individual clusters were observed between Abx-treated and untreated AF group (Fig 7D) Similarly separated clusters in the AB group lungs (Fig 7E) had dynamic enrichment/depletion in 249 Firmicutes, Fusobacteria and Proteobacteria after Abx-treatment.

250
Clearly separated clusters were evident between the Abx-treated and untreated FB group in both 251 niches (Fig 7F). Interestingly, the majority of taxa have increased abundance with Streptococcus being 252 the only taxa with decreased abundance after treatment in both BAL and lung in FB group. Overlapping 253 clusters between Abx-treated and untreated mouse microbiota in the BAL and lungs were only observed 254 in the AFB group (Fig 7G) wherein most of the identified taxa had reduced abundance; the only 255 exceptions were Streptococcus and Anaerococcus that were enriched after antibiotic treatment in both 256 niches.

363
Each mouse was weighed prior to challenge and every 24 hours after until harvest to monitor weight 364 change. Animals that lost more than 30% of their starting body weight, or exhibited severe signs of 365 morbidity, were euthanized for ethical reasons and recorded as having died on that day. Each group of 366 mice that were infected with a single agent were named after the pathogen as influenza virus (Flu Ctr) 367 and bacteria (Bact Ctr) only. Naïve (N) mice that had no treatments served to determine baseline 368 immune and microbiome information. Allergic mice that were infected with influenza (Flu) virus (23) 369 were referred to as the "Asthma and Influenza" (AF) group and those additionally infected with bacteria 370 were considered the "Asthma, Influenza, and Bacteria" (AFB) group. Allergic mice that were infected 371 with bacteria were referred to as the "Asthma and Bacteria" (AB) group.   Images were processed with Living Image software, version 4.5.5. Settings for image acquisition were a 383 one second photograph following by one minute luminescence measurement with the bining set at four.

Tissue harvest and bacterial enumeration 386
Tissues were harvested within a class 2A biosafety cabinet with strict adherence to aseptic 387 technique to ensure that samples were not contaminated by exposure to environmental agents.   Whole lung homogenate, whole blood, and BAL samples were serially diluted in sterile PBS and 400 10 µL of dilutions were plated on blood agar plates. Plates were incubated at 37°C with 5% CO 2 for 12-401 13 hours, and colony growth was enumerated. processed by QIIME (76). The closed reference mapping protocol of QIIME was used for OTU 453 assignment. Specifically, the read sequences were clustered into Operational Taxonomic Units (OTUs) 454 at 97% sequence similarity using the UCLUST algorithm (77). A representative sequence was then 455 selected from each OTU for taxonomic assignment using the Greengenes database (78) as the reference.

456
The alpha diversity estimates were calculated using the R Phyloseq package (79). Kruskal-Wallis non-457 parametric tests was used to test the significance of the diversity difference. The linear discriminant 458 analysis (LDA) effect size (LEfSe) method was used to test the significant difference of relative 459 abundance of taxa among groups (80).