Abstract
Streptococcus pneumoniae (pneumococcus) is one of the primary bacterial pathogens that complicates influenza virus infections. These secondary infections increase influenza-associated morbidity and mortality through a number of immunological and viral-mediated mechanisms. However, little is known about how specific bacterial genes contribute to post-influenza pathogenicity. Thus, we used genome-wide transposon mutagenesis (Tn-Seq) to reveal bacterial genes conferring improved fitness in influenza infected hosts. The majority of the 32 identified genes are involved in bacterial metabolism, including nucleotide biosynthesis, amino acid biosynthesis, protein translation, and membrane transport. We investigated five of the genes in detail: SPD1414, SPD2047 (cbiO1), SPD0058 (purD), SPD1098, and SPD0822 (proB). Single-gene deletion mutants showed slight growth attenuations in vitro and in vivo, but still grew to high titers in both naïve and influenza-infected murine hosts. Despite high bacterial loads in the lung and sustained bacteremia, mortality was significantly reduced or delayed with each of the knockouts. Reductions in pulmonary neutrophils, inflammatory macrophages, and select proinflammatory cytokines and chemokines were observed at discrete times after coinfection with these bacterial mutants. Immunohistochemical staining also revealed altered neutrophil phenotype and distribution in the lungs of animals coinfected with knockouts. These studies demonstrate a critical role for specific bacterial genes in driving virulence and immune function during influenza-associated bacterial pneumonia.
Author Summary Streptococcus pneumoniae (pneumococcus) is a common coinfecting pathogen that increases morbidity and mortality during influenza epidemics and pandemics. It is known that the strain, dose, and timing of bacterial coinfection influence the likelihood of severe pneumonia, but the specific bacterial genes that contribute to bacterial pathogenicity during influenza coinfection remain unknown. Using a genome-wide analysis, we identified the pneumococcal genes that exacerbate disease during influenza-bacterial coinfection. Most of these have a role in metabolism. To better understand their contribution to this lethal disease, we generated 5 mutants that lacked a single gene. The strains grew to high titers in the lungs and blood of both healthy and influenza-infected animals yet mortality was significantly reduced. In influenza-infected animals, there was also significantly lower inflammatory immune responses, and lung pathology. These important pneumococcal adaptations largely facilitate lethality during influenza-pneumococcal coinfection. Investigating whether similar metabolic adaptations are conserved among bacterial species that complicate influenza could yield broadly effective therapies that abrogate lethal post-influenza bacterial infections.
Introduction
Bacterial pathogens often complicate influenza virus infections, causing increased morbidity and mortality. Streptococcus pneumoniae (pneumococcus) is one of the leading pathogens causing death in influenza-infected hosts (1-3). During past pandemics, pneumococcal infection has presented as a risk factor for hospitalization and severe disease and has accounted for a significant proportion of influenza-related deaths (4-7). There is considerable genetic diversity within and between pneumococcal serotypes, and laboratory and clinical studies suggest that specific strains are preferentially promoted in influenza-infected hosts (8-10). Despite the clinical importance of this synergy, the bacterial factors that contribute to coinfection disease severity have yet to be determined in a systematic way.
During influenza A virus (IAV) coinfection with pneumococcus, bacteria are able to grow rapidly, viral burden increases, and significant inflammation amasses (Reviewed in references 11-16). The host-pathogen interplay is complex with numerous factors contributing to pathogen growth and host disease. Several studies have investigated the impact of viral virulence factors (17-25), bacterial virulence factors (23, 26), and host immune responses (Reviewed in references 11-15, 21, 27) to the pathogenicity of bacterial coinfections during influenza infection. Incidence of coinfection is, in part, a function of the detrimental effects that influenza virus infection has on key immune responses. Initial pneumococcal invasion and growth kinetics are regulated by virus induced depletion (28-31) or dysfunction (31-33) of alveolar macrophages (AMΦ), depending on the animal model (31). That is, the ratio of functional AMΦ and bacteria determines the initial bacterial growth phenotype and bacterial concentration increases when AMΦ become sufficiently limited in their capacity to phagocytose bacteria and, thus, provide sustained protection (30). Additional functional defects in AMΦ, neutrophils, and inflammatory macrophages are also present and likely contribute to sustained bacterial burden (31, 34-42). Several studies indicate that influenza-bacterial coinfection disease severity results from an immune ‘storm’, characterized by abundant immune cell infiltration and hyper-production of damaging pro-inflammatory cytokines (Reviewed in references 12-16, 21, 43). However, it remains unclear how specific bacterial species and/or strains influence these immune responses and coinfection pathogenicity.
Direct investigation into known pneumococcal virulence factors in influenza-infected hosts has provided some support that the bacteria need to adapt to a changing environment (22, 25). In the nasopharynx, bacterial sialic acid catabolism genes are important for enhanced bacterial growth during influenza virus infection (25). This may be a consequence of increased sialic acid availability mediated by viral and/or bacterial neuraminidases (NA)(25). Although viral NA can increase bacterial attachment to infected cells (18, 44), this does not seem to contribute significantly to bacterial growth (29). However, bacterial interactions with virus-infected cells do have a role in other aspects of the infection, including the viral rebound often observed post-bacterial coinfection (18, 29).
During primary pneumococcal infection, several genes have been identified as important regulators of disease severity (Reviewed in references 45-47). Systematic genomic screens have not yet been employed to assess pneumococcal adaptations during influenza coinfection. However, they were used to investigate which genes were necessary for growth of Haemophilus influenzae in the presence of influenza virus (48). Several influenza-induced metabolic adaptations of H. influenzae were identified, including changes in purine biosynthesis, amino acid metabolism, iron homeostasis, and cell wall synthesis (48). Genomic screens assessing pneumococcal adaptations in the context of other comorbidities have suggested similar alterations. For example, in hosts with sickle cell anemia, pneumococcal alterations in the TIGR4 strain included genes involved in complement function, iron acquisition, and purine biosynthesis (49). Given these important findings and the similar adaptations in bacterial metabolism under various host pressures, understanding how bacterial genes influence influenza-pneumococcal coinfection is critical and could identify mechanisms conserved across bacterial species amenable to targeting with therapeutics.
Here, we sought to identify specific pneumococcal genes that affect pathogenicity in influenza-infected hosts. To investigate this question, we used the genome-wide tool transposon insertion sequencing (Tn-Seq)(50) to reveal genotypes that confer bacterial fitness during viral coinfection compared to primary bacterial infection. This analysis identified 32 genes, most with metabolic functions, that contribute to pneumococcal fitness during influenza coinfection and do so in a time-dependent manner. To determine how select genes affect pathogenicity and host immune functions, we generated 5 single-gene deletion mutants (D39ΔcbiO1, D39ΔpurD, D39Δ1414, D39Δ1098, and D39ΔproB). The lethality of these knockout strains was significantly reduced in influenza-coinfected animals and significantly reduced or eliminated in naïve animals. This improvement in survival occurred despite high bacterial loads in the lungs and blood. However, select host immune responses were significantly suppressed, pulmonary consolidation was reduced, and pulmonary neutrophils had a more intact, functional morphology when pneumococci lacked these genes during influenza-coinfection. Taken together, these data indicate a critical role for pneumococcal metabolism in shaping host responses and disease severity during post-influenza bacterial pneumonia.
Results
Bacterial Adaptations During Pneumococcal Coinfection With Influenza
To identify which pneumococcal factors are required for growth during IAV infection, we employed a high-throughput, transposon sequencing (Tn-Seq) approach (50). We generated a pool of ∼50,000 mutants in the type 2 pneumococcal strain D39, then infected groups of mice with 75 TCID50 influenza A/Puerto Rico/34/8 (PR8) or mock control (PBS) followed 7 d post infection (pi) with 1e6 CFU of the mutant library. Lungs were harvested at 12 h or 24 h post-bacterial infection (pbi) and the bacteria were collected, genomic DNA was isolated and sequenced, and the relative abundance and fitness of mutants was calculated (details in Supplementary Information). In comparing the results from pre- and post-infection and from IAV- and PBS-infected animals, 17 genes conferred differential fitness at 12 h, and 23 genes had differential fitness at 24 h pbi. Of these, 8 genes were detected at both time points (Table 1, Fig 1).
The core set of genes identified at both 12 h and 24 h pbi are responsible for amino acid biosynthesis, nucleotide biosynthesis, protein translation, and membrane transport (Fig 1). Genes in the purine biosynthesis pathway comprised the largest number of genes (8 total; SPD0002 (dnaN), SPD0052, SPD0053 (purF), SPD0054 (purM), SPD0055 (purN), SPD0057 (purH), SPD0058 (purD), and SPD0059 (purE)) followed by ATP-binding cassette (ABC) transporters (5 total; SPD1098, SPD1099, SPD2047 (cbiO1), SPD2048 (cbiO2), and SPD1354 (putative)), protein translation (6 total; SPD0395 (efp), SPD1782 (ksgA), SPD0907 (hemK), SPD1130 (licD2), SPD1293, SPD1923), and proline biosynthesis (3 total; SPD0822 (proB), SPD0823 (proA), SPD0824 (proC)). Other genes included a putative membrane protein (SPD1090), carbon metabolism (SPD0723 (ripA), SPD1087, SPD1333 (putative), and SPD1468), and riboflavin metabolism (SPD0994).
Impaired Bacterial Metabolism Selectively Reduces Fitness In Vitro
To assess the differential fitness of genes predicted by Tn-seq, we generated 5 single-gene deletion bacterial mutants (D39ΔcbiO1, D39ΔpurD, D39Δ1414, D39Δ1098, and D39ΔproB) (Table 1, Fig 1, and Tables S2-S3). In vitro growth of the knockouts in ThyB was unaffected by the gene deletion with the exception of D39ΔcbiO1, which was significantly attenuated from 3 h to 8 h (Fig 2A). Consistent with this finding was the observed morphological changes of D39ΔcbiO1, which tended to be smaller and grow slower than wild-type (WT) D39 on blood agar plates. Interestingly, D39ΔcbiO1 was the only bacterial strain that had not lysed after 24 h in culture (Fig 2A). We examined decay kinetics in PBS to assess the fitness of knockout bacteria under conditions of metabolic starvation. The decay rates of knockout bacteria were similar to that of WT D39 with the exception of D39Δ1098 and D39ΔcbiO1 (Fig 2B). D39Δ1098 decayed rapidly and became undetectable 3 h before WT D39 (p=0.01) while D39ΔcbiO1 decayed more slowly and survived in culture 3 h longer than WT D39 (p<0.01) (Fig 2B). Although the decay rate of D39Δ1414 was similar to WT D39 (p=0.25), the rate increased at 9 h of culture and bacteria became undetectable 3 h earlier than WT D39 (Fig 2B). Growth of each of the knockout bacteria was reconstituted following metabolic starvation when cultures were supplemented with lung homogenate supernatants (s/n) from mock- or IAV-infected mice (Fig S1).
To assess the requirement of influenza virus-mediators for bacterial growth of the knockouts, we examined growth kinetics in lung homogenate s/n from naïve or IAV-infected mice (75 TCID50 PR8, 7 d pi) (Fig 2C-D). In both naïve and IAV-infected lungs, knockout bacteria titers were only slightly lower than WT D39 (p>0.05) after 6 hours of growth (Fig 2C-D). Specifically, after 6 h of growth in naïve lung s/n, WT D39 titers were 8.7 log10 CFU/ml, while titers of knockout bacteria ranged from ∼7.0 to 8.0 log10 CFU/ml (all p>0.05) (Fig 2C). In lung s/n from IAV-infected mice, titers of WT D39 were 8.3 log10 CFU/ml, while titers of knockout bacteria ranged from ∼6.7 to 8.0 log10 CFU/ml (all p>0.05) (Fig 2D).
Impaired Bacterial Metabolism Protects Against Virulence In vivo
Reduced Mortality
Mice were infected with 75 TCID50 PR8 or mock control (PBS) followed by 1e6 CFU WT D39 or one of the knockouts at 7 d pi to examine the effect of gene deletion on pathogenicity (Fig 3, Table S4). In PBS- and IAV-infected animals, infection with WT D39 resulted in 100% mortality by 72 h pbi and 48 pbi, respectively. In PBS-infected animals, mortality was reduced by 90-100% in 4 out of 5 knockouts and by 40% with D39ΔproB, compared to WT D39 (all p<0.01) (Fig 3C, Table S4). Correspondingly, weight loss at 48 h pbi with each knockout was significantly reduced compared to WT D39 (p<0.05) (Fig 3A). In IAV-infected animals, mortality was reduced by 40-90% in 4 out of 5 knockouts compared to WT D39 (all p<0.01) (Fig 3D, Table S4). Coinfection with D39ΔproB resulted in 100% mortality; however, the mean survival time was lengthened by 5 d (p<0.01) (Fig 3D, Table S4). In IAV-infected animals, weight loss was not significantly reduced compared to WT D39 (p>0.05) (Fig 3B).
Reduced Bacteria in the Lung and Blood
The growth capabilities of knockout bacteria observed in lung supplemented cultures in vitro (Fig 2C-D) were mirrored in in vivo growth kinetics (Fig 4A-B). Specifically, in both PBS- and IAV-infected animals, titers were lower than WT D39 in the lungs for each knockout bacteria at 4 h pbi (p<0.01) (Fig 4A-B, Table S4). By 24 h pbi, titers were similar to WT D39 in both the lungs and blood of PBS-infected animals (p>0.05) (Fig 4A, Fig 4C, and Table S4). However, at 24 h pbi in IAV-infected animals, each of the knockout bacteria grew to significantly lower titers than WT D39 in both the lungs (p<0.05) and blood (p<0.01) (Fig 4B, Fig 4D, and Table S4). Of note, D39ΔcbiO1 was not detected in the lungs of mock infected animals at 24 h pbi and was measured at low levels in 2 out of 5 mice at 48 h pbi and 1 out of 5 mice at 72 h pbi (Fig 4A). Despite attenuated growth in the lungs and reduced bacteremia at 24 h pbi, bacterial loads for each knockout remained high in the lungs and blood of coinfected animals at 24 h, 48 h, and 72 h pbi (Fig 4B, Fig 4D, and Table S4).
Similar Viral Load Kinetics
Similar to previous studies, viral loads rebounded following coinfection with each bacterial strain (Fig 4E, Table S4). Although the viral loads were not significantly different between the knockouts and WT D39 (p>0.05), there were some kinetic differences amongst the knockouts. The viral rebound during coinfection with D39ΔpurD and D39Δ1098 was delayed and the peak viral rebound in D39Δ1098 was slightly lower compared to coinfection with the other knockouts and WT D39 (Fig 4E, Table S4).
Altered Cytokine and Chemokine Responses
Because there was marked improvement in morbidity and mortality despite high pathogen burdens, we examined how infection with each of the knockouts affected cytokine and chemokine dynamics in the lungs (Fig 5, Fig S2-S4). In IAV-infected animals, there were significantly reduced levels of IFN-α, IFN-β, IL-6, KC, MIP-1β and GM-CSF at 24 h pbi during coinfection with knockout bacteria compared to WT D39 (p<0.05), with the exception of IFN-α during D39Δ1414 (p=0.95) and D39ΔpurD (p=0.17) coinfection (Fig 5, Fig S2-S4). IFN-α, IL-6, KC, and MIP-1β were not lower at 24 h pbi (p>0.05) with each of the knockout bacteria in PBS-infected animals (Fig 5, Fig S2-S4). IL-1α, IL-1β, MIP-1α, and TNF-α were reduced in both PBS- and IAV-infected animals at 24 h pbi with each knockout bacteria compared to WT D39 (p<0.05), except for D39ΔpurD in PBS-infected animals (p>0.05) (Fig 5, Fig S2-S4). Interestingly, at 4 h pbi in IAV-infected animals, IFN-α was elevated in D39Δ1414, D39ΔcbiO1, D39ΔpurD, and D39Δ1098 coinfections (p<0.01) and IFN-β was elevated in D39ΔcbiO1 coinfection (p<0.01) (Fig 5, Fig S2J and L). Minimal changes were detected in amounts of MCP-1, RANTES, IL-2, IFN-γ, IL-10, IL-12(p40), and IL-12(p70) during coinfection with any of the knockouts compared to WT D39 (Fig 5, Fig S3-S4).
Altered Cellular Responses
In accordance with the changes detected in pulmonary cytokines and chemokines, infection with knockout bacteria altered dynamics of select immune cells in the lungs (Fig 6 and Figs S5-S6). Most notably, in IAV-infected animals, neutrophils (Ly6Ghi) were significantly reduced at 4 h pbi during coinfection with D39Δ1098 (p<0.01) and D39ΔproB (p<0.05), and at 24 h pbi during coinfection with D39Δ1414, D39ΔcbiO1, D39ΔpurD, and D39ΔproB (p<0.05) (Fig 6B) compared to coinfection with WT D39. Inflammatory macrophages (IMΦs; Ly6G-, CD11chi, F4/80hi, CD11b+) were reduced at 4 h pbi during coinfection with D39Δ1414, D39ΔpurD, D39Δ1098, and D39ΔproB (p<0.05) (Fig 6D). Coinfection with D39ΔcbiO1 did not lead to reduced IMΦs at 4 h pbi (p=0.14), but did enduce a significant increase in IMΦs at 24 h pbi (p<0.01) (Fig 6D). In PBS-infected animals, neutrophils and IMΦs were not significantly different in any of the knockouts compared to WT D39 at 4 h pbi (p>0.05) (Fig 6A, Fig 6C). At 24 h pbi, only infection with D39ΔpurD (p<0.05) and D39ΔcbiO1 (p<0.05) resulted in altered neutrophils and IMΦs, respectively (Fig 6A, Fig 6C). There were minimal differences in the extent of AMΦ depletion (Ly6G-, CD11chi, F4/80hi, CD11b-) (Fig 6E-F) or T cell subset (Fig S6I-L) in PBS- and IAV-infected animals with knockout bacteria compared to WT D39.
Reduced Pathology and Altered Neutrophil Phenotype
Performing hematoxylin and eosin (HE) staining on the lungs of coinfected mice at 24 h pbi revealed extensive pulmonary consolidation in the lungs of mice coinfected with WT D39, as characterized by thickened septa and alveoli filled with a mixture of neutrophils, free bacteria, and proteinaceous exudates (Fig 7A). This was dramatically reduced during coinfection with D39ΔproB, D39Δ1098, and D39Δ1414 and absent in D39ΔcbiO1 and D39ΔpurD coinfection (Fig 7A). Immunohistochemical (IHC) staining showed that there was less bacterial antigen at 24 h pbi during coinfection with each of the knockout bacteria compared to WT D39 (Fig 7B), which mirrored the pulmonary bacterial loads (Fig 4B). In WT D39 coinfected mice, there was intracellular and extracellular bacterial antigen present throughout influenza-lesioned areas, including pervascular connective tissues, consolidated alveolar parenchyma, and the central hypocellular area of resolving lesions. Bacterial antigen was not detected in the resolving influenza lesions with any of the knockout bacteria, except for D39Δ1098 where few bacteria or antigen-positive macrophages were present. IHC staining also showed drastic neutrophil infiltration in WT D39 coinfection, and the resolving influenza lesions were surrounded by sharply demarcated hypercellular bands consisting of intact and degenerating neutrophils (Fig 7C). Neutrophils were reduced, although still abundant, in lungs of animals coinfected with D39ΔproB, D39ΔpurD, D39Δ1098. However, all cells were intact and occured in indistinct bands surrounding influenza lesions, clustered within alveoli peripheral to influenza lesions, or scattered throughout inflamed areas (Fig 7C). Animals coinfected with D39Δ1414 and D39ΔcbiO1 had the least neutrophils, which were morphologically intact and scattered throughout inflamed areas (Fig 7C). Neutrophils were rare or absent within the resolving influenza lesions in animals coinfected with each of the knockout bacteria.
Discussion
Pathogenicity during influenza-pneumococcal coinfection is influenced by several factors, including the viral and bacterial strain and doses, and the strength of the inflammatory response (Reviewed in references 11-16). Although viral and bacterial strain variability does influence coinfection severity (8-10, 51), how specific genetic factors contribute to pathogenicity has been relatively unexplored. Pneumococci, in particular, are highly adaptable and alter gene expression, protein production and modification, and metabolic functions to adjust to multiple host niches during infection (22, 52-66). In addition, its distributed genome is due to frequent recombination and horizontal gene transfer events (52), which facilitate its versatility. This is unlike the strategies employed by other bacteria like Staphylococcus aureus, another prominent influenza-coinfecting agent (1, 4, 5, 8, 67), where approximately 70% of the genome is conserved (68, 69). Given the genomic diversity and flexibility within pneumococcal strains, we sought to establish the genetic contribution to clinical phenotype of secondary bacterial pneumonia.
Here, using Tn-Seq as an unbiased approach to identify bacterial genes critical to pneumococcal survival in the IAV-infected host, we pinpointed 32 genes that alter bacterial fitness in a time-dependent manner (Table 1, Fig 1). By generating 5 single-gene deletion mutants of D39, we established that bacteria with altered metabolic capacities induce dramatically altered immune responses in IAV-infected hosts (Fig 5-7, S2-4, and S6). This reduction in immune activation, immune cell infiltration, and pulmonary damage, also occurred in a time-dependant manner (Fig 5-7, S2-4, and S6). These findings are provocative and consistent with other studies that have discovered time-dependent mechanisms during influenza-pneumococcal coinfection (29, 30, 70). Of note, none of the genes critical for growth in IAV-infected hosts were known bacterial virulence factors (45-47). This is likely because those factors affect disease equally regardless of viral infection status. The identified genes are mostly involved in bacterial metabolism, which supports findings that the metabolome of the lung is altered during influenza virus infection (71-74). Indeed, altered metabolic regulation has been observed during murine influenza virus infection (71, 72), in vitro influenza virus infection of primary human bronchotracheal epithelial (HBAE) cells (73, 74), and in pediatric patients infected with influenza virus (73). However, even with the knowledge that influenza virus induces metabolic changes and that pneumococci modulate metabolic pathways under host-specific pressures (49, 75), the specific collection of genes identified here was not intuitive.
Some common features did emerge when comparing our results with other studies. For example, adaptations specific to purine biosynthesis were previously found to occur in Haemophilus influenzae within influenza-infected hosts (48) and in pneumococci within hosts with sickle-cell disease (49). Here, eight genes with roles in purine metabolism were identified in our screen, including SPD0058 (purD) (76-78) (Table 1, Fig 1). This could indicate a common mechanism for bacterial adaptation in inflammatory environments. Specifically, in the influenza-infected host, purine biosynthesis is upregulated (71) and purine analogs can be used to reduce disease severity (e.g., by the antiviral T-705(79, 80)). However, the intermediates of purine biosynthesis become depleted following bacterial infection (48), rendering purine biosynthesis a critical function for coinfecting bacteria (48, 49). D39ΔpurD would rely heavily on purine scavenging, which could alter host immune cell function through competition for environmentally available purines and contribute to decreased lethality (Fig 3D) and immunopathology (Fig 7A) compared to WT bacteria in IAV-infected hosts.
Several of the genes identified here act in glutamate/glutamine biosynthesis, which suggests that these changes have a key role during influenza-pneumococcal coinfection. For example, the locus pinpointed as the main ABC glutamine/glutamate transporter in pneumococci is SPD1098/1099(81, 82), which was identified in our screen. Deletion of SPD1098 had the most profound improval on survival (90%) in coinfected animals (Fig 3D). In addition, SPD0822/0823 (proB/A), which act in proline biosynthesis, were identified in the screen and are downstream of glutamate metabolism (76-78). Deletion of SPD0822 (proB) did not lower coinfection pathogenicity, but did alter the time course of disease and delay mortality (Fig 3D). Interestingly, D39Δ1098 and D39ΔproB led to reduced neutrophils and iMΦ in influenza-infected hosts early in the infection (4 h pbi), but resulted in no changes in naïve hosts (Fig 6). In general, glutamine is utilized at a high rate by immune cells and is needed for optimal function of macrophages and neutrophils (83-88). In addition, pulmonary cells have increased dependence on glutamine during influenza virus infection (73). Thus, during influenza-pneumococcal coinfection, pneumococci and/or influenza-infected cells are in competition for glutamine with host immune cells. Here, altered competition for limited enivonmentally available glutamine may have reduced neutrophil infiltration and dysfunction, resulting in less morbidity and mortality in IAV-infected hosts (Fig 3D and Fig 5-7).
The function of some of the genes identified in our screen, including three other ABC transporters (e.g., SPD2047/2048 (cbiO1/2) and SPD1414), are less understood. In general, ABC transporters are important for pneumococcal virulence (52, 60, 89-92). Thus, it was expected that deleting these genes would reduce the growth capacity of D39. Futher understanding the specific impacts of each knockout mutant may provide insight into other respiratory changes that occur during influenza virus infection. For example, bacteria lacking the cbiO locus (putative cobalt transporter), previously identified by microarray analysis as necessary for pneumococcal pathogenicity (62), grow to significantly higher titers in IAV-infected animals than naïve animals (Fig 4A-D), suggesting that cobalt metabolism is modified by influenza virus infection. The TIGR4 analog of the oxalate/formate antiporter SPD1414 (SP1587) is important for bacterial survival in the lungs and blood (60, 93), cerebral spinal fluid (60), and nasopharynx (93). Here, deletion of SPD1414 did not result in lower bacteremia in naïve animals (Fig 4C), but did reduce bacteremia, neutrophil infiltration, and pulmonary damage in IAV-infected hosts the most dramatically (Fig 4D, Fig 6-7). It is unclear how each of these alterations contributed to the improved survival in IAV-infected hosts (Fig 3D).
Mortality during influenza-pneumococcal coinfection is typically associated with an exuberant immune response coupled with high pathogen loads in the lung and the blood (14-16, 34, 94-96). However, reduced inflammation can lessen severity even with sustained bacterial loads (36). Here, the attenuated growth of each of the knockout bacteria in IAV-infected animals was significant (0.8-2.1 log10 reduction compared to WT D39), but insufficient to account for the extreme reductions in mortality (up to 90%) (Fig 3D, Fig 4, Table S4). AMΦ, which dictate the initial pneumococcal invasion and growth kinetics during IAV infection (28-31), are not different during coinfection with each knockout bacteria compared to WT D39 (Fig 6F), except for D39ΔproB at 4 h pbi and D39ΔpurD at 24h pbi. This suggests that the observed differences in immune responses and disease outcome are not driven by the differences in bacterial loads. Reduced inflammation, specifically lowered type I IFNs (Fig 5, Fig S2J and L) during coinfection with knockout bacteria may have improved neutrophil function (32, 34, 35, 70, 97-100) and reduced epithelial cell death and lung permeability (101) and, thus, reduced coinfection pathogenicity (Fig 3D, Fig 5, and Fig 7). Moreover, reduced neutrophil infiltration and degredation (Fig 6B, Fig 7C) likely mitigated the damaging cytokine storm that is typically associated with IAV-pneumococcal pneumonia (Reviewed in references 12-16, 21, 43).
Here, there were not significant differences in viral load rebounds (Fig 4E) or contraction of the pulmonary CD8+ T cells (102) (Fig S6L) during coinfection with knockout bacteria despite reduced lung pathology, hypercytokinemia, and lethality (Fig 3D, Fig 5, and Fig 7). These results support our previous findings that the mechanisms underlying rapid bacterial growth are independent from those that influence the post-bacterial viral rebound and pathogenicity (29, 30). A direct correlation between survival and any single host immune response in the early stages of coinfection (0-24 h pbi) was not readily apparent. It is possible that unmeasured components and/or a cumulative effect influences lethality at later time points. These studies underscore the independent nature of pathogen growth and pathogenicity and illuminate the difficulty in reducing coinfection pathogenicity to a single variable.
Understanding how bacterial adaptations influence the development of pneumonia during influenza virus infections is important to effectively combat the disease. Here, we provide insight into the contribution of specific pneumococcal genes and to the regulatory host-pathogen dynamics that arise during the coinfection. Our findings highlight the critical role of influenza-mediated metabolomic shifts in inducing immune defects and promoting bacterial infection, and suggest that targeting a single pneumococcal gene or metabolite could be an effective intervention to abrogate bacterial pneumonia during influenza infection. Further dissecting bacterial adaptation may identify additional therapeutic targets that could be used to prevent or treat post-influenza bacterial infections.
Materials and Methods
Ethics Statement
All experimental procedures were performed under protocol O2A-020 approved by the Animal Care and Use Committee at St. Jude Children’s Research Hospital under relevant institutional and American Veterinary Medical Association (AVMA) guidelines and were performed in a biosafety level 2 facility that is accredited by the American Association for Laboratory Animal Science (AALAS).
Tn-Seq
Plasmid DNA harboring magellan6, a derivative of the Himar1 Mariner transposon, was purified from E. coli with the Qiagen mini plasmid preparation kit (Qiagen). Pneumococcal DNA was isolated by phenol/chloroform extraction and ethanol precipitation from an exponentially growing culture in ThyB media (30 mg/ml Todd-Hewitt Broth powder and 0.2 mg/ml yeast). In vitro, magellan6 transposition reactions were carried out with purified MarC9 transposase, 1 µg of pneumococcal target DNA and 1 µg of magellan6 plasmid DNA. Reactions were incubated for 1 h at 30°C, inactivated for 20 min at 72°C, ethanol precipitated and resuspended in gap repair buffer [50 mM Tris (pH 7.8), 10 mM MgCl2, 1 mM DTT, 100 nM dNTPs and 50 ng BSA]. Repair of transposition product gaps was performed with E. coli DNA ligase overnight at 16°C. Repaired transposition products were transformed into naturally competent pneumococcal strain D39. The following day, colonies were scraped off tryptic soy-agar (TSA) plates supplemented with 3% sheep erythrocytes and 200 mg/ml spectinomycin (TSA-Spec), pooled into libraries of approximately 50,000 transformants/library, split up into multiple starter cultures and stored at - 20°C.
Mice
Adult (6 week old) female BALB/cJ mice were obtained from Jackson Laboratories (Bar Harbor, ME). Mice were housed in groups of 5 in high-temperature 31.2cm x 23.5cm x 15.2cm polycarbonate cages with isolator lids. Rooms used for housing animals were maintained on a 12:12-hour light:dark cycle at 22 ± 2°C with 50% humidity in the biosafety level 2 facility at St. Jude Children’s Research Hospital (Memphis, TN). Prior to inclusion in the experiments, mice were allowed at least 7 days to acclimate to the animal facility such that they were 7 weeks old at the time of infection. Laboratory Autoclavable Rodent Diet (PMI Nutrition International, St. Louis, MO) and autoclaved water were available ad libitum. All experiments were performed under an approved protocol and in accordance with the guidelines set forth by the Animal Care and Use Committee at St. Jude Children’s Research Hospital.
Infectious Agents
All experiments were done using the mouse adapted influenza A/Puerto Rico/8/34 (H1N1) (PR8) and type 2 pneumococcal strain D39 variants. To generate the D39 knockout mutants, genomic DNA was isolated by phenol:chloroform extraction from an exponentially growing culture of WT D39 in ThyB media following genomic lysis (37°C, 10 min, (25 mM Tris-HCl, 10 mM EDTA, 50 mM glucose, 0.1 mg/ml RNase, 5 mg/ml deoxycholic acid, 5 mg/ml sodium dodecyl sulfate, and 8 u Proteinase K in molecular grade dH2O)). The same technique was used to isolate genomic DNA from S. pneumoniae (T4ΔHtrA) containing the erythromycin (ERM) resistance cassette. These were used as template DNA for gene SOEing PCR (103, 104). In brief, regions flanking the D39 locus targeted for deletion and containing overhangs complementary to the ERM resistance cassette (labeled regions A and B) were generated using the primer sets indicated in Table S2. Regions A, B, and ERM resistance cassette were amplified at the cycle conditions indicated in Table S2 (EasyA PCR kit (Agilent Tech) and BioRad T100 thermal cycler). Products were purified from a 1% agarose gel using the Zymoclean gel DNA recovery kit (Zymo Research). Regions A, B, and ERM resistance cassette were recombined by PCR (TaKaRa PCR kit (Clontech Laboratories)) using the primers and cycle conditions in Table S3. The recombined products were purified from gel as above and transformed into D39. Transformed bacteria were selected after overnight growth on TSA plates containing 1 µg/ml ERM (TSA-ERM). Infection stocks were grown in ThyB media containing 1 µg/ml ERM, and frozen in 12% glycerol stocks for animal infections. ERM resistance cassette insertion and target locus deletion were confirmed by PCR with the primer sets in Table S3.
Infection Experiments
The viral infectious dose (TCID50) was determined by interpolation using the method of Reed and Muench (105) using serial dilutions of virus on Madin-Darby canine kidney (MDCK) cells. The bacterial infectious dose (CFU) was determined by using serial dilutions on TSA (WT), TSA-ERM (knockouts), or TSA-Spec (Tn-seq) plates. Inocula were diluted in sterile PBS and administered intranasally to groups of 5 (for kinetics) or 10 (for bacteria collection and survival) mice lightly anesthetized with 2.5% inhaled isoflurane (Baxter, Deerfield, IL) in a total volume of 100 µl (50 µl per nostril). Mice were inoculated with either PBS or 75 TCID50 PR8 at day 0 then with the indicated CFU of D39 or knockout mutant (in 100 µl), 7 days later. Animals were weighed at the onset of infection and each subsequent day to monitor illness and mortality. Mice were euthanized if they became moribund or lost 30% of their starting body weight.
Lung and Blood Harvest for Bacteria Sequencing
Mice were euthanized by CO2 asphyxiation. Lungs were perfused with 10 ml PBS, aseptically harvested, washed three times in PBS, and placed on ice in 500 µl PBS. The post-perfusion fluid (mixture of blood and PBS) was plated immediately on TSA-Spec plates (150 µl/plate). The lungs were then enzyme digested with collagenase (1 mg/ml, Sigma), and physically homogenized by syringe plunger against a 40µm cell strainer. Cell suspensions were centrifuged at 4°C, 500xg for 7 min and the supernatant was plated on TSA-spec plates (100 µl/plate).
Bacterial Collection for Sequencing
Following infection, 500 µl of the inoculum was plated on TSA-Spec plates (100 µl/plate). Bacteria were collected from infected mice at 12 h pbi and 24 h pbi and plated on TSA-Spec plates (500 µl; 100 µl/plate). Later time points could not be examined due to insufficient numbers of live mice at 48 h pbi. For each mouse, ∼200 µl blood was plated upon harvest, and ∼500 µl of lung supernatant was plated following lung digestion (see above). Bacteria were incubated for 12 h at 37°C then collected in ThyB media and centrifuged at 4°C, 500xg for 10 min. The media supernatant was removed and the pellets were stored at −20°C.
Bacterial Fitness By Tn-Seq
Bacterial gene identification by Tn-Seq was performed as described previously (49, 50, 93). The samples at each of the three time points (pre-selection (inoculum, t1) and post-selection (after infection, 12 h (t2) or 24 h (t3) pbi), were sequenced in rapid run mode on an Illumina HiSeq 2000. For each insertion, the fitness Wi, was calculated by comparing the fold expansion of the mutant relative to the rest of the population with the following equation (106). where the mutant frequency at time 0 and harvest are Ni(t1) and Ni(t2,3), respectively. The expansion factor (d) accounts for bacterial growth during library selection. Additional details of the method are included in the Supplementary Information.
Lung Harvest for In vitro Kinetics
Mice were euthanized by CO2 asphyxiation. Lungs were aseptically harvested, washed three times in PBS, and placed in 500 µl PBS. Lungs were homogenized (Omni TH-01 with 5mm flat blade) and centrifuged at 4°C, 500xg for 7 min. Bacteria were grown at 37°C in 1.0 ml lung supernatants or in 1.0 ml PBS. A subset of PBS cultures were supplemented with 0.5 ml lung homogenate supernatant after 5 h of metabolic starvation (Fig S1). At each time point, 50 µl was removed, serially diluted in PBS, and plated on TSA (WT) or TSA-ERM plates. Bacterial titers were normalized to the total volume.
Lung Harvest for In Vivo Kinetics
Mice were euthanized by CO2 asphyxiation. Lungs were aseptically harvested, washed three times in PBS, and placed in 500 µl PBS. Lungs were enzyme digested with collagenase (1 mg/ml, Sigma C0130), and physically homogenized by syringe plunger against a 40 µm cell strainer. Cell suspensions were centrifuged at 4°C, 500xg for 7 min and the supernatants were used to determine the viral titers, bacterial titers, cytokine/chemokine levels (5 mice/group). Following red blood cell lysis, cells were washed in MACS buffer (PBS, 0.1 M EDTA, 0.01 M HEPES, 5 mM EDTA and 5% heat-inactivated FBS), counted with trypan blue exclusion using a Cell Countess System (Invitrogen, Grand Island, NY), and prepared for flow cytometric analysis as described below.
Lung and Blood Titers
For each mouse, viral titers were obtained using serial dilutions on MDCK monolayers, and bacterial lung titers and bacterial blood titers were obtained using serial dilutions on TSA (WT) or TSA-ERM (knockouts) plates, respectively.
Cytokines
Cytokines and chemokines were measured in lung supernatant by luminex (GM-CSF, IFN-γ, IL-1α, IL-1β, IL-2, IL-6, IL-10, IL-12(p40), IL-12(p70), KC, MCP-1, MIP-1α, MIP-1β, RANTES, and TNF-α) and ELISA (IFN-α,β). Prior to use, cell debris and aggregates were removed from lung supernatants by centrifugation at 4°C, 400xg. Milliplex magnetic bead cytokine/chemokine plates (Millipore) were prepared according to manufacturer’s instructions. Analysis was done using a BioRad BioPlex (HTF System) and Luminex xPonent software. Concentrations of duplicate samples were determined by construction of a standard curve for each analyte with a weighted 5PL method. ELISAs for IFN-α and IFN-β (PBL Assay Science) were prepared according to the manufacturer’s instructions. Plates were read at 450 nm and analyzed using elisaanalysis.com. Mean concentrations of duplicate samples were determined by construction of a standard curve with a 4PL regression. Absolute quantities of each cytokine/chemokine were calculated based on mean concentration of replicate samples normalized to the lung supernatant volume collected during tissue processing.
Flow Cytometric Analysis
Flow cytometry (LSRII Fortessa; Becton Dickinson, San Jose, CA) was performed on single cell suspensions after incubation with 200 µl of 1:2 dilution of Fc block (human-γ globulin) on ice for 30 min, followed by surface marker staining with anti-mouse antibodies: CD11c (eFluor450, eBioscience), CD11b (Alexa700, BD Biosciences), Ly6G (PerCp-Cy5.5, Biolegend), Ly6C (APC, eBioscience), F4/80 (PE, eBioscience), CD3e (PE-Cy7, BD Biosciences or BV785, Biolegend), CD4 (PE-Cy5, BD Biosciences), CD8α (BV605, BD Biosciences), DX5 (APC-Cy7, Biolegend or APC-e780, Affymetrix Inc) and MHC-II (FITC, eBioscience). The data were analyzed using FlowJo 10.4.2 (Tree Star, Ashland, OR) where viable cells were gated from a forward scatter/side scatter plot and singlet inclusion (Fig S5). Following neutrophil exclusion (Ly6Ghi), macrophages (MΦ) were gated as CD11chiF4/80hi with alveolar macrophages (AMΦ) sub-gated as CD11b- and inflammatory macrophages (iMΦ) as CD11b+. After macrophage exclusion, T cell populations were gated as CD3e+. The CD3e+ subset was subgated into CD8 T cells (CD3+CD8+CD4-DX5-) and CD4 T cells (CD3+CD8-CD4+DX5-). From the CD3e- population, natural killer (NK) cells were gated as CD3-DX5+ and dendritic cells (DCs) as CD3-DX5-. DC were further gated into three subsets of DC; CD11c+CD11b-, CD11c+CD11b+, and CD11c-CD11b+. The expression levels of MHC-II was used to confirm the identities and activation of MΦ and DC subsets. The absolute numbers of cell types were calculated based on viable events analyzed by flow cytometry and normalized to the total number of viable cells per sample.
Histology
Mice were euthanized by CO2 asphyxiation and lungs were inflated in situ via tracheal infusion with 10% neutral-buffered formalin solution (NBF; ThermoFisher Scientific, Waltham, MA), followed by continued fixation in NBF for at least 2 weeks before being embedded in paraffin, sectioned at 4 μm, mounted on positively charged glass slides (Superfrost Plus; Thermo Fisher Scientific, Waltham, MA), and dried at 60°C for 20 min. Tissue sections were stained with hematoxylin and eosin (HE) or subjected to immunohistochemical (IHC) staining to detect influenza antigen, pneumococcus, or neutrophils. For detection of these targets, tissue sections underwent antigen retrieval in a prediluted Cell Conditioning Solution (CC1; Cat# 950-124; Ventana Medical Systems, Indianapolis, IN) for 32 min on a Discovery Ultra immunostainer (Ventana Medical Systems, Tucson, AZ). The primary antibodies used in this study included: a polyclonal goat antibody raised against the HA glycoprotein of B/Florida/04/2006 (Yamagata lineage) influenza virus diluted 1:2000 (cat# I7650-05G, US Biologicals, Swampscott, MA); a rabbit polyclonal antibody to Streptococcus pneumoniae diluted 1:1000 (cat# NB100-64502; Novus Biologicals, Littleton, CO); and a rat monoclonal antibody to neutrophils (Ly6G6C) diluted 1:50 (cat# NB600-1387; Novus Biologicals, Littleton, CO). Binding of these primary antibodies was detected using OmniMap anti-Goat (#760-4647), anti-Rabbit (#760-4311), and anti-Rat (#760-4457) HRP (RUO) respectively (Ventana Medical Systems), with DISCOVERY ChromoMap DAB Kit (Ventana Medical Systems) as chromogenic substrate. Stained sections were examined by a pathologist blinded to the experimental group assignments.
Statistical Analysis
Significant differences in Kaplan-Meier survival curves were determined using the log rank test. In vitro growth/decay rates were analyzed by nonlinear regression of log10 values, and linear slopes were compared by analysis of covariance (ANCOVA). The remainder of statistical analyses were performed on linear values. Unpaired t-tests were done to analyze in vitro growth dynamics in ThyB and lung cultures. Analyses of variance (ANOVA) were performed using a Dunnett correction for multiple comparisons (to WT D39) to analyze in vivo differences including, lung and blood bacterial loads, viral loads, immune cells, cytokines, or chemokines (GraphPad Prism 7.0c). The confidence interval of significance was set to 95%, and p values less than 0.05 were considered significant.
Contributions
AMS, JAM, and JR conceived the experiments. AMS, AI, CB, and MJ generated the mutant libraries. DR, RC, and TvO completed the sequence analyses. AMS, APS, and LL generated the knockout bacteria and performed in vitro/in vivo experiments. AMS, APS, LL, and GH performed the data analysis. AMS and APS wrote the manuscript.
Acknowledgements
This work was supported by NIH grant AI100946 and AI125324 (AMS), PATRIC DBP (AMS, JR, JAM), ALSAC (APS, LCL, SW, RC, AI, CB, PV, DR, MDLJ, JAM, JR), NIH grant R25CA23944 (GH), and NIH grants R01AI110724, U01AI124302, and R21AI117247 (TvO).
References
- 1.↵
- 2.
- 3.↵
- 4.↵
- 5.↵
- 6.
- 7.↵
- 8.↵
- 9.
- 10.↵
- 11.↵
- 12.↵
- 13.
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.
- 20.
- 21.↵
- 22.↵
- 23.↵
- 24.
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.
- 38.
- 39.
- 40.
- 41.
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.
- 54.
- 55.
- 56.
- 57.
- 58.
- 59.
- 60.↵
- 61.
- 62.↵
- 63.
- 64.
- 65.
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.
- 85.
- 86.
- 87.
- 88.↵
- 89.↵
- 90.
- 91.
- 92.↵
- 93.↵
- 94.↵
- 95.
- 96.↵
- 97.↵
- 98.
- 99.
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵