Neisseria meningitidis is a narrow-host-range microorganism that colonizes the nasopharyngeal mucosae of approximately 10% of healthy subjects and only rarely causes life-threatening diseases, such as meningitis and sepsis. In industrialized countries, more than 60% of patients present with meningitis without septic shock (
5). The pathogenesis of meningococcal disease consists of several steps, including invasion of the bloodstream from the nasopharyngeal mucosa, survival in the blood, and entry into the central nervous system (CNS) by crossing the blood-brain barrier (
57). When
N. meningitidis massively replicates in the blood with concomitant endotoxin shedding, high levels of proinflammatory mediators are systemically released. In this case, fulminant meningococcal sepsis may occur even without induction of meningitis (
5,
57). In contrast, if bacteria multiply in the blood poorly, they may reach the CNS, with subsequent development of meningitis (
4,
57).
The long history of animal models of meningococcal disease dates back to 1907, when meningitis was induced in monkeys (
10). Presently, two rodent models (mouse and rat) and two routes of infection (intraperitoneal [i.p.] and intranasal [i.n.]) are employed to induce meningococcal disease. The i.p. model does not mimic the natural route of infection in humans but offers the advantage of inducing severe sepsis. The i.n. route is useful to analyze disease pathogenesis, but animals may develop lung infection before sepsis, in contrast to humans, who do not generally present with pneumonia. The model established in the rat is based on i.p. injection of neonatal animals with rat-passaged meningococci to increase bacterial virulence (
40), and it has largely been used in vaccine studies (
14,
41,
59). Mouse models of meningococcal disease are generally based on administration of an exogenous iron source to animals prior to infection in order to favor bacterial multiplication in the host (
17,
42). The i.p. mouse model was instrumental to assess protection from meningococcal challenge (
14,
33,
34). i.n. models have mainly used neonatal mice (
28,
38) to analyze disease pathogenesis and virulence of meningococcal isolates (
29,
37). Interestingly, meningococcemia was also induced by the i.n. route in adult mice superinfected with influenza A virus (
1). To overcome some shortcomings of traditional models, alternative strategies have also been explored (
8), including the use of transgenic mice expressing human CD46 (
18) or transferrin (
60). However, hardly any effort has been made to establish models of meningitis (rather than sepsis), and to our knowledge, no such a model has been developed in mice infected by the intracranic route.
N. meningitidis is able to obtain and synthesize nutrients essential for its survival in the different environments within the human host during infection. A genome-wide analysis of virulence genes required for systemic meningococcal infection in the rat showed that about half encode enzymes involved in metabolism and transport of nutrients (
51). There is evidence that
l-glutamate uptake from the host is critical for meningococcal infection in both cell and animal infection models (
30,
51). We have demonstrated that GltT, an ABC-type transporter for
l-glutamate, is essential for meningococcal growth/survival in infected cells (
30) and that invasive isolates hyperexpress
gdhA, encoding the NADP-specific
l-glutamate dehydrogenase (
35). Indeed, meningococci are naturally auxotrophic for
l-glutamate and use it as a carbon (and nitrogen) source by supplying the tricarboxylic acid cycle with 2-oxoglutarate when glucose or lactate is limiting (
30,
35). Due to this auxotrophy, host
l-glutamate is essential for the biosynthesis of several amino acids and the antioxidant glutathione (
http://www.genome.ad.jp/dbgetbin/get_pathway?org_name=nme&mapno=00251 ).
In this study, we have developed a novel model of meningococcal meningitis based on intracisternal (i.cist.) infection of adult mice. The model was used to assess the virulence of a mutant strain deficient in the l-glutamate transporter GltT. Here, evidence that the GltT mutant is attenuated compared to the wild-type strain is provided, suggesting that uptake of l-glutamate is crucial for the development of meningococcal meningitis in mice.
MATERIALS AND METHODS
Bacterial strains and growth conditions.
The meningococcal strains used in this study are the serogroup C 93/4286 isolate and its isogenic mutant 93/4286ΩgltT, which is devoid of the ABC-type l-glutamate transporter GltT. The 93/4286 strain, belonging to the ET-37 hypervirulent lineage, was kindly provided by Novartis Vaccine and Diagnostics, Siena, Italy. Meningococci were cultured on blood agar medium (Oxoid) or gonococcus (GC) (Oxoid) agar/broth supplemented with 1% (vol/vol) Polyvitox (Oxoid) at 37°C with 5% CO2. When needed, erythromycin (Sigma-Aldrich) was added to a final concentration of 7 μg ml−1. Escherichia coli strain DH5α was used in cloning procedures. This strain was grown in Luria-Bertani (LB) (Oxoid) medium. To allow plasmid selection, LB medium was supplemented with ampicillin (50 μg ml−1).
DNA procedures, plasmids, and transformation of meningococci.
High-molecular-weight genomic DNA from
N. meningitidis strains was prepared as previously reported (
6). DNA fragments were isolated by using acrylamide slab gels and recovered by electroelution as described before (
39). Amplification reactions were as follows: 45 s of denaturation at 94°C, 45 s of annealing at 65°C, and 60 s of extension at 72°C for a total of 30 cycles. Reactions were carried out in a Perkin-Elmer Cetus DNA Thermal Cycler 480 using oligonucleotides NMB1965-1 and NMB1965-2 (
30). The DNA of strain 93/4286 was used as a template. Southern blot hybridizations were carried out according to standard protocols (
39).
32P labeling of DNA fragments was performed by random priming using the Klenow fragment of
E. coli DNA polymerase I and [α-
32P]dGTP (3,000 Ci mmol
−1) (
39).
The
Neisseria-E. coli shuttle plasmids pDEX and pDEΔNMB1965 were as previously described (
30,
35). The NMC1937 open reading frame of strain 93/4286, coding for the permease of the GltT transporter, was genetically inactivated by single crossover using plasmid pDEΔNMB1965. The resulting mutant was named 93/4286ΩgltT. Transformation experiments were performed by using 0.1 to 1 μg of plasmid DNA as previously described (
11). Transformants were selected on GC agar medium supplemented with erythromycin. Gene inactivation was demonstrated by Southern blot hybridization using a 482-bp-long NMB1965-specific
32P-labeled probe (
30). Primer synthesis and DNA sequencing were performed by Ceinge Biotecnologie Avanzate s.c.ar.l., Naples, Italy. DNA sequence analysis was carried out by using the GeneJockey Sequence Processor software (Biosoft).
Mice.
Eight-week-old female outbred CD1 mice weighing 24 to 30 g were obtained from Charles River (Calco, Italy). Animals were allowed to settle in the new environment for 1 week before the experiments were performed. All animal experiments were approved by the local ethics committee (document no. 754/03, 12.9.03) and were carried out according to institutional guidelines.
Mouse model of meningococcal meningitis.
Mice were infected by the i.cist. route by a technique described by Koedel et al. to induce meningitis by
Streptococcus pneumoniae (
20). In vivo pilot experimentation was carried out using laboratory-grown
N. meningitidis. Because the results were not compelling enough, subsequent experiments were all performed with mouse-passaged bacteria to increase meningococcal virulence. Briefly, animals were lightly anesthetized (50 mg/kg ketamine and 3 mg/kg xylazine). Approximately 10
7 CFU of bacteria in a total volume of 20 μl were inoculated by hand-puncturing the cisterna magna of mice using a 29-gauge needle (Artsana, Italy). Meningococci were recovered 24 h later by homogenizing the brain with a screen mesh in 1 ml of GC medium. Passaged bacteria were grown to early exponential phase in GC broth to an optical density of 0.7 at 600 nm, corresponding to approximately 7 × 10
8 CFU ml
−1. Bacteria were stored frozen at −80°C in GC broth supplemented with 10% (vol/vol) glycerol (Carlo Erba) until use.
Prior to infection, passaged bacteria were thawed at room temperature, harvested by centrifugation for 5 min at 1,800 × g, and resuspended in fresh GC broth containing iron dextran (5 mg kg−1). Viable counts were performed on blood agar plates to determine the exact number of CFU. Approximately 5 h before infection, animals were injected i.p. with iron dextran (250 mg kg−1). Bacteria were injected by the i.cist. route as described above. Animals were monitored for seizures due to inoculation.
Animal survival, clinical parameters, and CFU counts.
Different bacterial doses ranging from 105 to 107 CFU per mouse were used to inoculate animals (n = 4 to 12) by the i.cist. route with mouse-passaged wild-type or GltT-deficient meningococci. Control mice were inoculated with GC broth. Every day throughout the whole experiment, animals were monitored for clinical symptoms (i.e., ruffled fur, hunched appearance, hypothermia, weight loss, lethargy, or moribund). Body weight and temperature were measured by using a digital balance (Acculab) and a thermometer (Greisinger Elettronic), respectively. Rodents were humanely killed before reaching the moribund state. Survival was recorded for a week.
To determine the number of wild-type meningococci in organs, animals (n = 16) were infected with 5 × 105 CFU/mouse and sacrificed at different time points (6, 12, 24, and 30 h) after infection. To compare virulence of wild-type versus GltT-deficient bacteria, two groups of mice (n = 30/group) were infected with 107 CFU/mouse and sacrificed at 6, 24, and 48 h after challenge for organ collection. All moribund mice were humanely killed. Blood was withdrawn by cardiac puncture before sacrifice and added to a tube containing 3.8% sodium citrate. Brain, spleen, and liver were excised and homogenized in 1 ml of GC medium. Viable counts were determined by plating 10-fold dilutions onto blood agar plates.
Mouse competition experiments.
Frozen culture stocks of mouse-passaged 93/4286 (erythromycin-sensitive) and 93/4286ΩgltT (erythromycin-resistant) strains were thawed, centrifuged, and resuspended in fresh GC broth with iron dextran (5 mg kg
−1) at the appropriate dilutions. Strains were mixed at a 1:1 ratio, and mice (
n = 9) were infected i.cist. with equivalent numbers of wild-type and mutant bacteria. Each animal received 2 × 10
7 total bacteria in a volume of 20 μl. Mice were killed at 6, 24, and 48 h after infection. Blood, brain, spleen, and liver were removed at each time point and treated as described above. Samples were plated onto blood agar plates with and without antibiotic selection to distinguish between strains. The competitive index (CI) was calculated as (CFU
mutant/CFU
wild type)
output/(CFU
mutant/CFU
wild type)
input, where the output represents the viable bacteria recovered from a target organ at a certain time point after infection and the input is the bacterial mixture in the initial inoculum. A CI of <1 indicates decreased growth (and fitness) of the 93/4286ΩgltT mutant compared to the wild type (
2,
3). When no bacteria were recovered from one or more mice, it was assumed that at least 1 CFU had been isolated. CI values are represented as means ± standard errors of the means (SEM).
Histological and immunofluorescence analysis.
For histological analysis, animals were inoculated with 10
7 CFU of bacteria from either mouse-passaged wild-type or GltT-deficient strains and sacrificed when moribund. Mice injected with GC medium were used as controls. Brains were excised, fixed in formalin for 24 h, and then embedded in paraffin according to standard procedures. Brains were sectioned along a coronal plane (
31). Tissue blocks were cut in 5-μm sections, stained with hematoxylin-eosin according to standard techniques, and examined by using routine light microscopy. In both mouse-passaged wild-type and GltT-deficient strains, the extent and degree of inflammatory infiltrates in the brain were evaluated in meningeal, intraparenchimal, and ventricular tissue areas (Table
1). Immunofluorescence was performed to distinguish between extracellular and intracellular meningococci in the brains of infected mice. All incubations were carried out at room temperature for 60 min unless stated otherwise. After being washed in phosphate-buffered saline (PBS), samples were incubated with polyclonal rabbit antibodies raised against whole meningococci (anti-
N. meningitidis 6121; ViroStat) and then stained with anti-rabbit immunoglobulin G conjugated with tetramethylrhodamine isothiocyanate (TRITC) (Dako). To detect intracellular bacteria, TRITC-stained tissues were permeabilized for 10 min with 0.25% saponin (Sigma-Aldrich) in PBS. Samples were incubated with anti-
N. meningitidis antibodies (in PBS and 0.25% saponin) and then treated with fluorescein isothiocyanate (FITC) (Dako)-conjugated anti-rabbit immunoglobulin G diluted in PBS and 0.25% saponin as previously described (
48,
53). Primary and secondary antibodies were diluted 1:500 and 1:200, respectively. Successful permeabilization was verified using a control antibody (MAB318 [Chemicon], 1:300) specific for an intraneuronal cytoplasmic target (tyrosine hydroxylase) followed by Alexa Fluor-conjugated anti-mouse antibodies (Molecular Probes, 1:1,000) (see Fig. S1 in the supplemental material). DAPI (4′,6′-diamidino-2-phenylindole) (Sigma-Aldrich; 1:1,000) staining was performed to reveal nuclei. Extra- and intracellular bacteria were visualized using a Zeiss Axioskop2 Plus fluorescence microscope (Zeiss).
Statistical analysis.
Differences in survival of mice inoculated with the wild type or the mutant were analyzed by using the Fisher exact test. The two-tailed Student
t test was employed to analyze differences in body weight and temperature between different animal groups (
P < 0.05). CFU counts in different organs and time points were represented as numbers of bacteria isolated from single mice, and means ± SEM for each animal group were calculated. Differences in bacterial loads between mice infected with the wild type and the mutant were determined with the Mann-Whitney U test (
P < 0.05), (
45).
DISCUSSION
In the present study, we have established a model of meningococcal meningitis in outbred adult mice based on i.cist. inoculation of bacteria. To our knowledge, no other model of meningococcal meningitis has been developed in mice infected i.cist., although the i.cist. route of infection has been explored to establish models of meningococcal meningitis in both the rat (
19,
55) and the rabbit (
56). As the highest rates of meningococcal disease are in young children, adolescents, and young adults (
12,
13), immunocompetent 8-week-old animals were employed rather than neonatal or infant subjects. Outbred (CD1) mice were preferred to inbred strains due to their cost-effectiveness and because they better mimic the natural variation in infection occurring in human population. We chose the i.cist. (
20) rather than the intracranic (
7) route of infection, because the former ensures complete delivery of the inoculum directly into the cisterna magna, thereby facilitating bacterial replication in the CSF. As shown by mouse survival, clinical parameters, and histology, i.cist. inoculation was functional at inducing the disease. Large doses (∼10
7 CFU) of group C
N. meningitidis were necessary to cause systemic infection and death of 50% of mice in comparison with
S. pneumoniae, where the 50% lethal dose was approximately 10
2 CFU (
7). In contrast, the use of lower i.cist. inocula (∼10
5 CFU) supported active bacterial replication in the CSF but resulted in clearance from the bloodstream. Overall, the bacterial doses used in this study were smaller with respect to the inocula (10
8 to 10
9 CFU) required to induce meningococcal disease in neonatal mice injected via the i.n. route (
29). Mouse passage was a key step to develop the model. Indeed, microorganisms recovered from the brains of infected mice were capable of replicating in the CSF and blood of naïve rodents. In contrast, systemic survival of the laboratory strain was severely hampered. Interestingly, both laboratory and mouse-passaged strains induced meningitis with histopathologic features mimicking those of the disease in humans, suggesting that replication of meningococcal strains (even those that are poorly virulent) may be facilitated in the CSF due to CNS immunodeficiency (
46,
47). Not only did the mouse-passaged isolate grow efficiently in the CSF, but it also persisted in the spleen and liver. During the hypoferremic phase of neisserial infection, most of heme-derived iron remains associated within liver ferritin (
26). As meningococci can obtain iron from ferritin (
22), the liver may represent a target organ for meningococcal replication. Altogether, the i.cist. model is effective at inducing both primary meningitis and disseminated meningococcal disease compared to i.p. or i.n. models, which are generally less suitable to study meningitis, as animals may die of sepsis before meningitis is established.
By using our murine model of meningitis, we tested the virulence of a strain deficient in the l-glutamate transporter GltT. Data on the survival and clinical parameters of rodents infected with the mutant suggested virulence reduction, but statistical significance was not compelling.
Analysis of meningococcal loads over time clearly proved that the GltT mutant was attenuated in the model. In contrast to wild-type bacteria, the GltT-deficient strain was cleared systemically the day after infection. The mutant did not exhibit active replication into CSF but could still be isolated from the brains of infected animals at 48 h after challenge, yet again suggesting CNS immunodeficiency (
46,
47). Despite persistence of the mutant in the CNS, meningeal inflammation was limited or absent in comparison to that caused by the wild-type strain, indicating that
l-glutamate transport plays a role in establishing the disease in the CNS. As assessment of virulence by standard methods (i.e., 50% lethal dose calculation) is a relatively crude and insensitive approach that may fail at evidencing the contribution of a given factor to pathogenicity (
3), we decided to use mixed infections to determine the degree of virulence attenuation of the GltT-deficient strain compared to the wild type. Indeed, reduction of virulence was proven in mixed meningococcal infections, where the mutant was clearly hampered compared to the parental strain. During disease progression, the GltT-deficient strain showed a decreasing capability to adapt to different body sites in the host. At 48 h after infection, CI values were comparable in each mouse compartment analyzed, indicating that the wild type outcompeted the mutant strain in a time-dependent and organ-independent manner. Attenuation of the GltT-deficient strain in the peripheral compartments was consistent with recent work demonstrating that a knockout GltT mutant had reduced survival in human blood and was attenuated in a systemic mouse model of infection (
27).
N. meningitidis is able to efficiently utilize only a few compounds as energy sources, including lactate, pyruvate, glucose, and maltose. Glucose and lactate are the predominant carbon sources in blood and at mucosal surfaces, respectively (
24,
25). In the intracellular milieu, where levels of glucose are low, the main energy sources for meningococci are pyruvate, lactate, and certain amino acids, such as
l-glutamate (
16,
52). The availability of
l-glutamate is considered critical for meningococcal infection in both cell (
27,
30) and animal infection (
27,
51) models and is also instrumental in preventing oxidative injury, as
l-glutamate is the precursor of glutathione (
52). In this study, we demonstrated that
l-glutamate uptake from the murine host is also required for bacterial replication and survival in the CSF, suggesting that the
l-glutamate may be important in meningitis both for supporting meningococcal growth in the CNS and in preventing bacterial damage due to oxidative stress. Interestingly, the levels of
l-glutamate in the brain and CSF are strongly increased in animals and humans suffering from bacterial meningitis (
15,
49) and correlate with disease severity (
50). It has been hypothesized that during meningitis
l-glutamate and other excitatory amino acids are released by both macrophages derived from blood monocytes and microglial cells during meningitis (
50) and that release of these amino acids may result in membrane depolarization and calcium influx, leading to energy failure and neuronal cell death (
32). These effects are mediated by the activation of the
N-methyl-
d-aspartate receptor complex. Therefore, it is believed that
l-glutamate release contributes to neuronal damage during bacterial meningitis and that
N-methyl-
d-aspartate receptor antagonists may be of therapeutic use (
21,
44). In this perspective, the use of
l-glutamate analogues to both halt meningococcal replication in the CSF and impede
l-glutamate neurotoxic effects may represent an attractive therapeutic strategy against bacterial meningitis.