Bacteremia and penetration of the blood-CSF barrier is necessary for development of the overwhelming majority of cases of meningitis. However, in addition to meningitis caused by spread of bacteria by the hematogenous route, meningitis has also been reported following infection with other bacteria as a complication of traumas, surgical procedures, and developmental malformations or as secondary infections to chronic ear infections. Regardless of the route of entry of bacteria, whether by hematogenous spread or by nonhematogenous routes, the host defenses within the SAS are inadequate to control infection. Classical bacterial meningitis is predominantly a leptomeningitis, with both the inflammatory response and bacteria largely contained within the SAS and with little or no involvement of the dura mater or the underlying brain.
DISCUSSION
In the present study, we investigated the nature of the in vitro interactions between cells derived from the human meninges and
Neisseria lactamica, which has also been associated with intracranial inflammation, like the closely related bacterium
Neisseria meningitidis. It is generally accepted that
N. lactamica associates primarily with cells of mucosal epithelia (
5,
22,
30,
53,
55), but significantly, we now demonstrate that
N. lactamica also shows a predilection for adhering to human meningeal cells. Moreover, the dynamics of association were similar to those observed for
N. meningitidis, although the overall levels of association with
N. lactamica were generally lower. In a previous study, pili were identified as the major ligand that mediated adherence of
N. meningitidis to meningeal cells (
32). Pilin (
pilE) genes are present in a wide range of
Neisseria species (
3), and analysis of
pilE loci from different
Neisseria species has shown the presence of two distinct structural classes: one class (class II) includes the pilin genes from
N. lactamica and some strains of
N. meningitidis, and the other (class I) contains gonococci and the remaining
N. meningitidis strains (
2). Expression of class I or class II pili by different strains of
N. meningitidis has been shown not to influence adherence to meningeal cells (
32), and since
N. lactamica also expresses class II pili, it is unlikely that the class of pilus expressed accounts for the observed differences in levels of adherence of this bacterium compared with
N. meningitidis. It is possible that adherence is influenced by differences between the species in expression of the pilin-associated protein PilC, which is present in
N. meningitidis but absent in
N. lactamica (
44,
45). However, other potential adhesins, such as the Opa and Opc proteins (
16,
61), which are both present in
N. meningitidis but absent in the
N. lactamica species used in this study, may also contribute to the observed differences in adherence between the two species. In addition, the fact that challenge with a saturating MOI of
N. lactamica, compared with the same concentrations of
N. meningitidis, still resulted in lower levels of
N. lactamica association with meningioma cells than
N. meningitidis at each time point suggests that a difference in host cell receptors used by these species and/or receptor density may also influence adherence.
The range of bacterial concentrations (ratio of MOI of bacteria to cells of 0.002:1 to 2,000:1) used to infect meningioma cells showed a correlation with the concentration of bacteria found in the CSF during natural infection. Although
N. lactamica and
N. meningitidis can be cultured from CSF, estimates of bacterial numbers in untreated patients with meningitis are unreliable even if care is taken to optimize sample collection and culture procedures (
7). In addition, reported numbers are likely to be underestimates that do not take into account bacterial cell death occurring in the CSF. In the case of
N. meningitidis, the levels of meningococcal LPS in the CSF accurately reflect the bacterial growth rate and the ability to release LPS (
7). Moreover, quantification of LPS content in meningococci showed that LPS at 100 ng/ml was equivalent to approximately 10
8 bacteria (
57). The study of Brandtzaeg et al. (
8) reported median LPS levels in the CSF of patients with meningitis at 2.5 ng/ml (range, 0.025 to 500 ng/ml), and this converts to approximately 2.5 × 10
6 CFU/ml (range, 2.5 × 10
4 to 2.5 × 10
8 CFU/ml). Thus, these median values are in general agreement with our concentrations of meningococci (10
2 to 10
8 CFU) used to infect meningioma cells. Although to our knowledge there are no published numbers for
N. lactamica recovered from the CSF of patients with meningitis, it is possible that large bacterial numbers could occur through direct inoculation from surgery or from other trauma.
An important finding from the present study was that cells derived from the meninges were involved in the innate host defense response to
N. lactamica. Meningioma cells challenged with this organism secreted a specific subset of proinflammatory (IL-6), chemoattractant (IL-8, MCP-1, and RANTES), and growth-factor-related (GM-CSF) cytokines. This profile of cytokine secretion was also observed following infection with
N. meningitidis (
13,
26), suggesting that
N. lactamica and
N. meningitidis share similar modulins capable of stimulating the meninges. However, although the levels of IL-8, MCP-1, and GM-CSF secretion induced by
N. lactamica and
N. meningitidis were essentially similar for each cytokine measured, some differences were observed in IL-6 and RANTES secretion.
N. lactamica induced lower levels of IL-6 secretion by meningeal cells than
N. meningitidis. Interestingly, OM from both bacteria were poor stimulators of IL-6 production, suggesting that components of the bacteria other than OM modulins are largely responsible for induction of this cytokine. This conclusion is consistent with our previous study, which demonstrated that OM and also pure LPS from meningococci had a minimal effect on IL-6 secretion (
13). Recent studies have shown that both immunoglobulin A (IgA) protease and peptidoglycan fragments, which are both secreted by pathogenic
Neisseria (
20,
39), up-regulate IL-6 production by peripheral blood mononuclear cells. However,
N. lactamica does not produce IgA protease (
44,
45), and it is possible that the observed induction of IL-6 secretion is due to peptidoglycan release, although the involvement of additional uncharacterized secreted components cannot be excluded (
49). In the case of RANTES, it was of particular interest that while low concentrations of both bacteria induced similar levels of secretion, high concentrations of
N. meningitidis significantly down-regulated chemokine production during infection. This was also observed with the highest concentration of
N. lactamica during infection, but the effect was less pronounced compared to that with
N. meningitidis, and such differences may be important contributions to the pathogenesis of infection caused by meningococci.
Knowledge of the modulins expressed by
N. lactamica and
N. meningitidis is important for understanding how both organisms are likely to interact with human meningeal cells. In the present study, the phenotype of the meningococcus was similar to that of strains isolated from the CSF of patients with meningitis (
19,
52) in expressing capsule, pili, LPS, and the major outer membrane proteins PorA, PorB, Opa, Opc, and Rmp. To our knowledge, no information is available on the antigens expressed by
Neisseria lactamica isolated from the CSF of patients with meningitis. However,
N. lactamica does express pili, LPS, PorB, and Rmp but not capsule, PorA, or Opc. Recently, comparative genomic studies of
N. lactamica and
N. meningitidis have demonstrated further differences between the two bacteria (
44,
45). Conserved DNA sequences encoding virulence-associated genes were identified in
N. meningitidis that were absent in
N. lactamica, including those genes encoding capsule synthesis, secreted Frp and RTX toxins, PilC, and IgA1 protease, as well as several genes whose products are homologous with virulence-associated proteins found in other bacteria. In addition, several genes encoding enzymes associated with metabolic pathways, transporter proteins, integral membrane proteins, and lipoproteins were present in
N. meningitidis and absent in
N. lactamica (
44,
45). Interestingly, an in vitro study of gene expression in
N. lactamica and
N. meningitidis on contact with epithelial cells demonstrated changes in expression of 285 genes in
N. lactamica and 347 genes in
N. meningitidis, with 167 genes common to both organisms (
30). Dissimilarities in gene expression between the bacteria, identified by both comparative genomics (
44,
45) and following host cell interaction (
30), may account in part for differences in the pathogenicity of
N. lactamica and
N. meningitidis, particularly with respect to the ability of the latter to invade the nasopharyngeal mucosa, survive in the blood, and penetrate the blood-CSF barrier.
Based on the data in the present study, a model by which
N. lactamica induces intracranial inflammation can be proposed. Despite the obvious genetic and phenotypic differences between the bacteria, following entry into the SAS,
N. lactamica is able to proliferate in the CSF, adhere to meningeal cells, and induce an inflammatory response similar to that observed with
N. meningitidis. The interactions of
N. lactamica, bacterial products, and, to a lesser extent, released OM induce the secretion of IL-6, IL-8, MCP-1, RANTES, and GM-CSF by leptomeningeal cells. This pattern of secretion is consistent with the elevated levels of these and other cytokines found in the CSF of patients with meningitis caused by
N. meningitidis and other bacteria (
12). Roles for IL-8 in the in vivo chemotaxis of polymorphonuclear leukocytes (PMNL) into the CSF, abetted by up-regulation of cell adhesion molecules on blood vessel endothelium, a consequence in part of RANTES secretion, have been demonstrated (
1,
23). However, the ability of large numbers of
N. lactamica cells to down-regulate RANTES production to some extent suggests that
N. lactamica may influence the ability of PMNL to invade the SAS, thereby allowing bacterial persistence in the CSF. Moreover, down-regulation of RANTES was even more pronounced following
N. meningitidis infection, suggesting that this bacterium shows a greater capacity for manipulating the host PMNL response. Clearance of bacteria in the SAS is also dependent on MCP-1 secretion, which results in monocyte accumulation in the SAS, with GM-CSF acting as a maturation factor (
27). Although our data suggest that lower levels of IL-6 production by the meninges in response to
N. lactamica may lead to a reduction in leukocytosis and induction of the fever response, this is likely to be outweighed by the proinflammatory cytokine production by infiltrating PMNL and monocytes and resident macrophages.
During the course of leptomeningitis, significant cell and tissue injury is observed in the SAS. Both
N. lactamica and
N. meningitidis have been reported to kill blood vessel endothelial cells in vitro (
21), exacerbating further entry into the CSF of bacteria and inflammatory cells. However, leptomeningitis is an acute, compartmentalized inflammatory response, and both the bacterial and inflammatory cell exudate are largely contained within the SAS. In particular, the inability of
N. lactamica to invade meningeal cells, a property shared with
N. meningitidis (
32), suggests that the pia mater provides a barrier to direct penetration of either bacterium to the sub-pial space and brain cortex below. However, a significant finding from the present study was that whereas meningeal cells remained viable on prolonged challenge with
N. lactamica, cell death was induced by
N. meningitidis and the mechanism was overwhelming necrosis with no significant apoptosis. Our observations on the lack of an apoptotic phenotype following meningococcal challenge are consistent with the reports from Wells et al. (
64) and Robinson et al. (
49), who demonstrated that meningococci induced elevated gene expression of antiapoptotic factors in meningioma cells and a small but significant resistance effect to staurosporine-induced apoptosis. Thus, the ability to kill meningeal cells suggests that meningococci and components of the pathogen may be able to interact with the underlying glia limitans superficial to the brain cortex, which is consistent with astrocyte activation that is often observed during bacterial meningitis (
28). It is possible that this sequence of events may result in a higher degree of neurological damage and death from
N. meningitidis meningitis compared with infection caused by
N. lactamica.
In summary, the interactions of N. lactamica with cells derived from the meninges share many similarities with N. meningitidis, suggesting that when Neisseria species enter the CSF the innate response of the human meninges is remarkably conserved. However, it is likely that differential expression of modulins between the bacteria contributes to the observed differences in pathogenic potential, such as cytokine regulation and induction of cell death, and consequently may influence the eventual prognosis for the patient with leptomeningitis. Identifying the nature of the bacterial components involved may suggest new targets for therapeutic intervention during infection.