We developed a model of sequential influenza A virus (IAV)–Neisseria meningitidis serogroup C (Nm) infection in BALB/c mice. Mice infected intranasally with a sublethal IAV dose (260 pfu) were superinfected intranasally with Nm. Fatal meningococcal pneumonia and bacteremia were observed in IAV-infected mice superinfected with Nm on day 7, but not in those superinfected on day 10. The susceptibility of mice to Nm superinfection was correlated with the peak interferon-γ production in the lungs and decrease in IAV load. After Nm challenge, both IAV-infected and uninfected control mice produced the inflammatory cytokines interleukin (IL)-1 and IL-6. However, IL-10 was detected in susceptible mice superinfected on day 7 after IAV infection, but not in resistant mice. This model of dual IAV–Nm infection was also used to evaluate the role of bacterial virulence factors in the synthesis of the capsule. A capsule-defective mutant was cleared from the lungs, whereas a mutant inactivated for the crgA gene, negatively regulating expression of the pili and capsule, upon contact with host cells, retained invasiveness. Therefore, this model of meningococcal disease in adult mice reproduces the pathogenesis of human meningococcemia with fatal sepsis, and is useful for analyzing known or new genes identified in genomic studies.
Meningococcal diseases are life-threatening in humans, and may occur as sporadic cases or outbreaks worldwide [1,2]. Epidemiological data indicate a strong association between the incidence of influenza and subsequent increases in the incidence of meningococcal diseases [3–7] and of other bacterial respiratory infections [8,9]. Invasive meningococcal infections, mainly bacteremia and meningitis, only occur in humans, the only known reservoir, in which Neisseria meningitidis is present as a commensal organism in the upper respiratory tract . Most of our knowledge concerning the pathogenesis of meningococcal infection has been derived from clinical investigations. The pathophysiologic process begins with colonization of the respiratory tract. This is followed by invasion of the blood vessels to induce bacteremia and then the crossing of the blood–brain barrier to reach the subarachnoidal spaces, inducing meningitis . However, N. meningitidis is not a primary pathogen, and the host factors involved in the pathogenesis of invasive infection are largely unknown . Identification of the mechanisms by which adhesion occurs and putative invasive bacterial determinants, particularly the capsule, interact with host tissues is hampered by the lack of a reproducible animal model mimicking the various steps of the infectious process, from respiratory tract colonization to invasion of blood [12,13]. Neonatal and infant mouse models have been shown to reproduce meningococcal respiratory infection followed by bacteremia, but adult mice are resistant to respiratory challenges [14,15].
In this study, we used the epidemiological association between influenza and meningococcal infection to develop a mouse model. We increased the susceptibility of the mice to bacterial infection through a previous influenza infection.
We evaluated the susceptibility of mice infected intranasally with influenza A virus (IAV) to superinfection with a serogroup C N. meningitidis strain or its isogenic derivatives, affected in capsule synthesis.
Materials and methods
Influenza A virus
A stock of the mouse-adapted strain, A/Scotland/20/74 (H3N2) , made from lung homogenates in 30% glycerol and stored at −80°C, was used to prepare inocula for mouse infection. We administered the virus intranasally to a mouse and recovered it 4 days later from the infected lungs. Virus titers were determined from the supernatants of lung homogenates centrifuged at 700×g for 15 min, by standard plaque assay, 72 h after the infection of Madin–Darby canine kidney cells.
The Streptococcus pneumoniae serotype 14 strain Pn40  was cultured in nutrient broth for pneumococci (Sanofi-Diagnostic Pasteur, Marnes la Coquette, France) or on blood agar (trypticase soy agar supplemented with 5% horse blood) (BioMérieux, Marcy l'Etoile, France).
N. meningitidis, clone 12 (Nm), is a derivative of strain 8013, a serogroup C, class 1 strain. It has the phenotype P+, Opa−, PilC1+/PilC2+. Nm was cultured on GCB agar medium (Difco) containing the supplements described by Kellog et al.  or in brain heart infusion broth (Difco). The CrgA mutant NM98-3, harboring an inactivated crgA allele, and the capsule mutant NM00-4, obtained by inactivating the siaD gene encoding polysialyltransferase, have been described elsewhere [19,20].
For challenge experiments in mice, bacterial inocula from subcultures in broth supplemented or not with the appropriate antibiotic were adjusted to the required density by dilution in phosphate-buffered saline (PBS) (Sigma, Saint Quentin Fallavier, France).
Five-week-old female BALB/c mice (Charles River, France) were kept in a biosafety containment facility, in filter-topped cages with sterile litter, water and food, according to institutional guidelines. They were challenged at the age of 6 weeks. The experimental design was approved by the Institut Pasteur Review Board.
Respiratory infection was induced by intranasal administration of 50 µl of viral or bacterial inoculum in mice lightly anesthetized with sodium pentobarbital (Sanofi, Santé Animale, Libourne, France). Bacterial counts were determined in mice killed at the times indicated from the number of colony-forming units (cfu).
Statistics were performed by comparing the geometric means in groups of five mice in bacterial challenge experiments, and in groups of three mice in other assays, by Student's t test.
Groups of three mice superinfected with Nm at day 7 after previous IAV infection were killed by intraperitoneal injection of 300 mg kg−1 sodium pentobarbital, either 3 h or 24 h after Nm challenge. A catheter connected to a container of 2% formaldehyde was inserted into the trachea via a ventral incision in the neck. They were simultaneously exsanguinated by retro-orbital puncture. The thorax was opened, and the lungs were immediately inflated in situ via the tracheal cannula, with 10% formaldehyde in PBS. The lungs were then removed and immersed in 10% formaldehyde for 8 days. They were processed for embedding in paraffin, and stained with hematein and Giemsa stain. To detect Nm in the lung tissues, we carried out immunohistochemistry with a specific anti-serogroup C antibody. In brief, deparaffinized lung sections were immersed in citrate buffer and treated with proteinase K. Endogenous peroxidase activity was blocked with H2O2, and a rabbit anti-N. meningitidis serogroup C antibody (specific for the capsule polysaccharide antigen) was added, at a dilution of 1:50. The sections were incubated for 1 h at room temperature, washed with PBS, and a biotinylated goat anti-rabbit IgG antibody (Dako EO432), diluted 1:500, was added for 45 min. The sections were then treated with horseradish peroxidase-conjugated streptavidin (Dako PO397) and the reaction was revealed by amino ethyl carbazole/H2O2. The preparations were then counterstained in hematoxylin. In addition, blood smears from the pulmonary blood vessels were prepared from three mice superinfected with Nm for 24 h and Gram-stained.
On predetermined days after IAV infection or after Nm superinfection, cell-free lung homogenates (centrifuged at 1800×g) and sera were collected and stored at −80°c before testing by ELISA (Quantikine M, R&D Systems Europe, Abingdon, UK) for the quantification of immunoreactive interferon-γ (IFNγ), and interleukins (IL)-1, 4, 6, 10, 12, and 13. The rat anti-mouse IFNγ neutralizing monoclonal antibody XMG1.2 or the isotype control IgG1 K, rat monoclonal antibody R3-34 (both from Pharmingen) were used for in vivo assays in IAV-infected mice. Recombinant IL-10 (R&D Systems Europe) was used in one experimental challenge with Nm.
Assessing the IAV/Nm model of sequential respiratory infections
Mice were challenged intranasally with serial 1:5 dilutions of a titrated IAV suspension in PBS as previously described . The mean 50% lethal dose was approximately 1000 plaque-forming units (pfu). The lowest sublethal IAV dose that reproducibly induced pneumonia that spontaneously resolved within 10 days was 260 pfu (Fig. 1). Mice convalescing from IAV were tested for susceptibility to respiratory superinfection with Nm or S. pneumoniae, before or 7, 14 or 21 days after IAV infection with 260 pfu. There was a transient phase of susceptibility to Nm and S. pneumoniae superinfections 7 days after IAV infection, but not thereafter, suggesting a transient, non-specific, flaw in the innate immune system (Table 1). We have shown that S. pneumoniae Pn40, serotype 14, does not induce pneumonia in immunocompetent mice, and that leukopenia is a major factor in susceptibility to respiratory challenge with this strain . Polymorphonuclear leukocytes also play a crucial role in primary immune defenses against N. meningitidis. Under our experimental conditions, mice responded to both IAV and subsequent bacterial respiratory infections by a doubling of polymorphonuclear blood counts within 48 h of each challenge. Strong recruitment of polymorphonuclear leukocytes occurred in lung, and blood samples showed phagocytosed meningococci (see Section 3.3).
|Inoculum (log10 cfu)||Bacterial load in the lungs (log10 cfu ml−1)|
|Day after IAV infection|
|N. meningitidis (7.1)||<1||7.3±0.7||<1||<1|
|S. pneumoniae (3.5)||1.9±0.8||7.2±0.5||3.2±0.3||2.7±0.3|
|Inoculum (log10 cfu)||Bacterial load in the lungs (log10 cfu ml−1)|
|Day after IAV infection|
|N. meningitidis (7.1)||<1||7.3±0.7||<1||<1|
|S. pneumoniae (3.5)||1.9±0.8||7.2±0.5||3.2±0.3||2.7±0.3|
Mean±S.E.M. for five mice at 72 h after challenge.
Meningococcal pneumonia and bacteremia
Mice infected intranasally with a standardized inoculum of 260 pfu of IAV were superinfected intranasally, 7 or 10 days later (at the stage of viral pneumonia or at the stage of viral clearance, respectively, see Fig. 1), with 7 log10 cfu of Nm (Fig. 2). In IAV-infected mice, superinfected with Nm on day 7, the lungs were colonized (with an increase in bacterial load of approximately 1 log10 cfu after 48 h) and hemorrhagic lobar pneumonia lesions were detected after as little as 24 h. About half of the mice died within 48 h. Bacteremia was detected after 3 h, reaching 4 log10 cfu ml−1 at 48 h. No clinical signs of brain disease were recorded and cultures from cerebrospinal fluid samples taken from mice with bacteremia were negative. Nm inocula with bacterial densities below 6.5 log10 cfu did not reproducibly induce colonization of the lungs in all mice, and bacteremia was detected only in mice with lung bacterial counts ≥7 log10 cfu at 24 h. In IAV-infected mice challenged with Nm on day 10, the respiratory inoculum was cleared within 48 h and no bacteremia was detected, as in control mice without previous IAV infection (Fig. 2).
Macroscopically, meningococcal respiratory superinfection in mice 7 days after previous IAV infection induced acute lobar pneumonia with large foci of alveolitis. The microscopic examination of bronchial sections at 24 h after Nm challenge showed localized lesions of the epithelium in which Nm clusters were stained with the specific antibody (Fig. 3A). This suggested that the bacteria adhered to and probably crossed the epithelial barrier at localized points of damage due to IAV infection. In the parenchyma, intense inflammation of the alveolar septa was observed with infiltrating polymorphonuclear leukocytes and the presence of cell-associated bacteria, suggesting phagocytosis of Nm by polymorphonuclear and monocytic cells (Fig. 3A′, B). This lesions were seen as early as 3 h after Nm challenge in all lung sections. At 24 h, perivascular infiltration of leukocytes was detected in several lung sections. In one lung section, we observed meningococci accumulating on the wall of a blood vessel causing the inflammatory reaction in the adjacent endothelium (Fig. 3C), and blood smears from pulmonary blood vessels revealed bacteremia with the presence of extracellular as well as intracellular Gram-negative diplococci in the cytoplasm of polymorphonuclear leukocytes (Fig. 3D), consistent with positive blood cultures (Fig. 2).
In this experimental model mild IAV infection did not induce leukopenia and mice recovered spontaneously. We therefore considered the possibility that transient cytokine dysregulation during the course of viral infection might act as a factor of susceptibility to bacterial respiratory superinfection in mice previously infected with IAV. We tested this hypothesis by evaluating the production of Th1-type (IFNγ, IL-1, and IL-12) and Th2-type (IL-4, IL-6, IL-10, and IL-13) cytokines in the lungs and sera of IAV-infected mice. The main cytokine response to IAV infection detected in the lungs of BALB/c mice was the production of IFNγ, which peaked on day 7 and returned to basal values on day 10 (Fig. 4). IL-1β was also detected, peaking on day 2 and progressively decreasing thereafter. Moderate production of IL-6 and IL-10 was detected from day 2 to day 10. The other cytokines tested were not detected. No significant amounts of the cytokines tested (<10 pg ml−1) were detected in serum. The correlation between the stage at which levels of production of the proinflammatory cytokine IFNγ were maximal during IAV pneumonia and the stage of maximal susceptibility of mice to bacterial superinfection with Nm or S. pneumoniae (Table 1) was surprising. It suggested a deleterious effect of the inflammatory reaction in the lungs on innate immunity to bacterial infection. We tested this hypothesis by administering the rat anti-mouse IFNγ neutralizing monoclonal antibody XMG1.2  or the isotype control IgG1 K rat monoclonal antibody R3-34 to IAV-infected mice on day 7, intranasally, at a dose of 30 µg per mouse, 1 h before Nm infection with 7.2±0.5 log10 cfu. This anti-IFNγ neutralization led to an increase in bacterial load at 24 h in both the lungs (>8 vs. 7.2±0.3 log10 cfu ml−1) and the blood (3.6±0.6 vs. 2.4±0.7 log10 cfu ml−1), suggesting that if virus-induced inflammation increased the susceptibility to the bacterial superinfection, this effect could not be attributed to direct effects of IFNγ.
Cytokine responses were further tested 24 h and 48 h after challenge with Nm, in the lungs of mice infected 7 or 10 days previously with IAV and in uninfected controls. Nm infection had no effect on IFNγ production. All mice with previous IAV infection, and uninfected controls, had detectable IL-1β and IL-6 responses 24 h after Nm superinfection. An IL-10 response was detected only in Nm-infected mice infected with IAV 7 days previously (P<0.01). We therefore investigated the effect of IL-10 on the susceptibility of mice, without previous IAV infection, to intranasal challenge with Nm. We administered 8.08 or 8.34 log10 cfu of Nm per mouse (in two independent experiments), intranasally in combination with 0.5 or 2.5 ng of recombinant mouse IL-10, corresponding to twice and 10 times, respectively, the concentration of IL-10 detected in susceptible mice (Fig. 4). IL-10 alone did not render the mice more susceptible to Nm, as no infection was detected in cultures from blood and lungs at 24 and 48 h.
Evaluating the role of Nm virulence determinants in the IAV–Nm dual infection model
IAV-infected mice were superinfected on day 7 with Nm or one of its isogenic mutants impaired in expression of genes for major virulence factors, the capsule and the regulatory protein CrgA. Ability to colonize the lungs was evaluated from changes in bacterial load in the lungs and the invasiveness of the strain was evaluated from changes in bacterial load in the blood. Comparative growth/survival kinetics for various Nm derivatives (Fig. 5) showed that the crgA mutant colonized the lungs more effectively, inducing higher levels of subsequent bacteremia, than the parent strain. As expected, the capsule mutant displayed an impaired ability to colonize the lungs and to invade the blood.
The association of influenza and bacterial respiratory superinfection has been well documented [7–9]. Various changes in respiratory epithelial cells have been described, including a loss of ciliated cells and the hyperplasia of Clara cells, associated with secondary bacterial pneumonia. The local recruitment of monocytes and polymorphonuclear leukocytes, which may be induced by the direct cytopathic effects of viral infection of these target cells, has been extensively studied [21–23].
In this study, we show that mild IAV infection may transiently modify immune defenses in the respiratory tract, rendering the host susceptible to pulmonary superinfection with serotype 14 S. pneumoniae or serogroup C N. meningitidis. These strains were not virulent in uninfected mice or mice that had recovered from IAV infection. At the stage of virus-induced susceptibility to bacterial superinfection, mice infected with this low-density IAV inoculum displayed a normal polymorphonuclear cell response to bacterial infection, suggesting that acute inflammatory cytokines were produced, as reported in mouse models [24,25] and in human studies . As mild IAV infection in this model did not induce leukopenia, and resolved spontaneously, we thought that the transient dysregulation of cytokine profiles during the course of the viral infection might affect the susceptibility to bacterial respiratory superinfection of mice previously infected with IAV. We tested this hypothesis by analyzing the production of Th1- and Th2-type cytokines , in the lungs and sera of IAV-infected mice. In our model of mild IAV infection, a cytokine response, mostly of the Th1 type, was detected in the lungs but not in serum, confirming previous reports [25,28].
IAV-infected mice were susceptible to Nm on day 7 but not on day 10 after IAV infection. Nm colonized the lungs and induced inflammatory pneumonia, with an intense influx of polymorphonuclear leukocytes and monocytes, followed by bacteremia. This invasive pneumonia was lethal in 48–72 h. Thus, this model of dual IAV–Nm infection mimicked the pathogenesis of human meningococcal disease . However, no meningeal infection was detected by culture of spinal fluid (positive cultures from the brain tissue or fluid would not be specific because Nm circulated in brain blood vessels due to the intense bacteremia). Therefore, this model of meningococcal disease in adult mice closely resembled meningococcemia with fatal sepsis.
The observation that mice with previous IAV infection became susceptible to Nm superinfection on day 7, correlating with peak of IFNγ production in the lungs, but recovered their primary resistance by day 10, when IFNγ was no longer detected, seems paradoxical, because IFNγ activates the phagocytic and bactericidal functions of macrophages and polymorphonuclear leukocytes. However, as IFNγ production was correlated with IAV replication, the phagocytic cells may have undergone viral-induced impairment of their antibacterial functions  at the stage of maximal IFNγ detection. The increase in bacterial loads observed after treatment with a neutralizing anti-IFNγ monoclonal antibody demonstrated that IFNγ-induced inflammation was not directly responsible for the susceptibility of mice to Nm superinfection.
We checked the effects of IAV-induced inflammation on the cytokine response of mice to Nm challenge. IAV-infected and naive mice responded to Nm challenge by producing IL-1 and IL-6, which are produced in large amounts in meningococcal disease [1,29]. IL-10 was produced by susceptible IAV-infected mice challenged on day 7, but only a weak response was detected in IAV-infected mice on day 10 and in controls. The anti-inflammatory cytokine IL-10 plays an important role in determining the severity of meningococcal disease [1,29]. IL-10 production may play a major role in this mouse model of dual IAV–Nm infection. However, the co-administration of mouse recombinant IL-10 with Nm inoculum in normal mice did not lead to infection, suggesting that other changes are involved in susceptibility to Nm after IAV infection. Further investigations are required at the cellular and molecular levels to identify the mechanisms by which IAV induces the transient changes in innate immunity that render the host susceptible to meningococcal superinfection.
This model of Nm invasive infection demonstrated certain cellular changes associated with the higher susceptibility to superinfection of IAV-infected mice. It was also useful for evaluating the role of specific virulence factors in Nm pathogenesis. The absence of negative regulation by CrgA of the expression of the pili and of the capsule upon contact with host cells  may be correlated with an increase in the persistence of bacteria on the respiratory epithelia and blood invasion, probably due to higher bacterial resistance to antibacterial blood components. The capsule-defective mutant was cleared rapidly. It did not colonize the lungs and was unable to invade the blood, confirming the major role played by the polysaccharidic capsule of Nm in resistance to host innate immunity. This demonstrates the value of this mouse model for analyzing meningococcemia. Other bacterial factors, and other genes identified in genomic studies  could also be evaluated for virulence and immunogenicity in this model of Nm infection in adult mice, which reproduces the main pathogenic features of meningococcemia occurring in influenza outbreaks.
We thank Prof. Claude Hannoun, for constant support in our experimental approach, Dr. Geneviève Milon and Dr. Daniel Scott-Algara for their advice in analysis of the cytokine responses, and Prof. Xavier Nassif, Prof. Philippe Sansonetti and Dr. Jean-Pierre Bouvet for fruitful discussions on the relevance of the experimental model. We would also like to thank Maryse Tardy-Panit, Marek Szatanik, and Huot Khun for their excellent technical assistance. This work was supported by the Institut Pasteur. A.A. holds a fellowship from the Fondation pour la Recherche Médicale.
influenza A virus
Nm mutant affected in contact-regulated gene A
capsuledefective mutant of Nm