Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Meningitis is an inflammation of the membranes enclosing the brain and spinal cord. Meningitis may be caused by viral, fungal, or parasitic infection, but is frequently caused by bacteria capable of penetrating the barriers protecting the brain. The barrier function of the blood–brain barrier (BBB) and the blood–cerebrospinal fluid barrier (BCSFB) largely excludes peripheral immune cells and plasma proteins in normal physiological conditions, which led to the concept of the central nervous system (CNS) as a site of immune privilege. The lack of immediate and sufficient bactericidal activity in the cerebrospinal fluid (CSF) allows bacteria to multiply almost as efficiently as in vitro, to reach titers of up to 109 colony-forming units (CFU)/mL and to spread over the entire surface of the brain, the spinal cord, and along penetrating vessels. This explains that prior to the antibiotic era, bacterial meningitis almost always led to death.

To gain access to the CNS, bacteria must (a) colonize the host, (b) invade the bloodstream or cause focal infection in the vicinity of the brain, (c) survive in the bloodstream, and (d) cross the BBB or the BCSFB (Fig. 1). Bacterial meningitis can arise from bacteremia if pathogens colonizing the nasopharyngeal and intestinal mucosa invade the bloodstream or by continuous infection from nearby foci, e.g., sinusitis, mastoiditis. It may also follow a breach of the brain barriers by a foreign body (e.g., cerebrospinal fluid shunt, cochlear implant), a neurosurgical procedure, or trauma.

Fig. 1
figure 1

Selected mechanisms that contribute to the pathogenesis of bacterial meningitis and the development of brain injury

Full size image

Once a pathogen has reached the CNS, a cascade of events follows until the full symptomatic manifestations of bacterial meningitis. These include the induction of cytokines and chemokines; activation of inflammatory mediators such as nitric oxide (NO), reactive oxygen species (ROS), or matrix metalloproteinases (MMP); recruitment of white blood cells to the site of infection; and cytotoxic events. In this chapter, the pathogenesis of bacterial meningitis and its neurological sequelae are discussed.

2 Pathogenesis of Bacterial Meningitis

2.1 Bacterial Invasion of the Host and Penetration of the Blood–Brain Barrier

Figure 2 shows the pathogenic steps involved in the development of bacterial meningitis.

Fig. 2
figure 2

Pathogenetic steps involved in the development of bacterial meningitis: (i) colonization of the nasopharynx and invasion of blood capillaries via crossing of the mucosal barrier, the epithelia and the vascular endothelium; (ii) bacteremia, survival in the blood and crossing of the blood–brain barrier (BBB); and (iii) bacterial multiplication in the CSF, induction of inflammation, BBB breakdown, and invasion of blood-derived neutrophils into the subarachnoid and ventricular space

Adherence of bacteria to the mucosal epithelium and colonization are affected by phase variation. Bacteria cross the mucosal barrier through or between epithelial cells. Once in the bloodstream, expression of a sufficiently thick capsule is necessary to protect the bacteria from circulating antibodies, complement-mediated bacterial killing, and neutrophil phagocytosis. Bacteria penetrate the BBB by (i) paracellular passage, (ii) transcellular passage, or (iii) invasion within white blood cells during diapedesis (intracellular pathogens). Neutrophilic pleocytosis occurs along a chemotactic gradient mainly created by chemokines locally expressed in the brain. Matrix metalloproteinases (MMP), such as MMP-8 and MMP-9, facilitate the process of extravasation by degrading extracellular matrix components of the brain microvasculature

Full size image

2.1.1 Bacterial Colonization of the Host

Colonization of the host is a common first step of the major bacterial pathogens that cause meningitis (Table 1). S. pneumoniae, N. meningitidis, and H. influenzae type B colonize the nasopharynx and are transmitted from person to person by the respiratory route. Injury of the functional integrity of the respiratory tract mucosa by viral infections [2, 3] and physical damage to the mucosa [4] may increase the risk of invasive bacterial disease.

Table 1 Most common bacterial organisms causing meningitis in humans according to broad age categories. Relative frequency of occurrence in studies performed in different European countries are given as percentage
Full size table

To colonize the nasopharyngeal mucosa, a pathogen first has to evade mucosal defense mechanisms like ciliary clearance [5–10], secretory immunoglobulin A (IgA) [11–13], and lysozyme [14–17]. Secondly, it has to prevail against other organisms in the aerobic environment of the upper respiratory tract [18–20]. Finally, it must adhere to the mucosal epithelium [10, 21–25].

Group B streptococcus (GBS) (S. agalactiae), E. coli, and L. monocytogenes colonize the gastrointestinal tract and are transmitted by the oral, vaginal, or fecal–oral route. They mainly cause meningitis at the vulnerable extremes of age and in immunocompromised persons. In neonatal meningitis, colonization of the intestinal tract of the infant follows similar principles as outlined for nasopharyngeal colonization [26–34]. The food-borne pathogen L. monocytogenes primarily affects immunocompromised patients and pregnant women and is transmitted to the fetus through the placenta or to the newborn during delivery [35, 36].

2.1.2 Invasion of the Bloodstream

To enter the bloodstream at the site of colonization, bacteria penetrate the mucosal barrier formed by epithelial cells and the lining of the blood vessels formed by the vascular endothelium (Fig. 2). Depending on the pathogen, the paracellular or transcellular route across the epithelial and endothelial layers is preferred.

Most known mechanisms by which meningeal pathogens invade human epithelial and endothelial cells via the transcellular route take advantage of the ubiquitous clathrin-mediated cellular endocytosis mechanism [37–39]. This then leads to the uptake into vacuoles that may either recycle back to the cell surface, fuse with lysosomes, or transit to the basolateral cell surface where the bacteria are released out of the cell [40–42].

Actin rearrangements further enhance bacterial internalization. Some transmembrane proteins of the epithelial cell that are targeted by the bacterial pathogen and initiate endocytosis are polymeric immunoglobulin receptor (pIgR) [43, 44], carcinoembryonic antigen-related cell adhesion molecule (CEACAM) proteins [45–47], and epithelial cadherin (E-cadherin) [48, 49].

Paracellular migration of bacteria across the epithelial barrier is facilitated by toll-like receptor (TLR)-dependent downregulation of tight junction protein expression [50] or extracellular degradation of junctional components mediated by bacterial enzymes or bacteria-bound plasmin [51–54].

The final barrier between epithelial cells and mesenchymal cells is the epithelial basement membrane. Meningeal pathogens may degrade the epithelial basement membrane by secreting proteolytic enzymes [51, 55], abuse the host’s plasminogen–plasmin system [53, 56–58], and cross the basement membrane in phagocytic cells [59, 60]. Finally, attachment to extracellular matrix (ECM) components is important in bacterial infection pathogenesis, and ECM adhesins have been identified in most meningeal pathogens (e.g., PavA of S. pneumoniae [21], Opc of N. meningitidis [61], fimbriae of H. influenzae [62], and ScpB of GBS [31]).

2.1.3 Intravascular Survival

Once in the bloodstream, meningeal pathogens are immediately exposed to the host’s immune defense. Clearance of S. pneumoniae, H. influenzae, and N. meningitidis strongly relies on antibody-mediated opsonization followed by activation of the complement system. S. pneumoniae and H. influenzae are then taken up by host leukocytes and killed intracellularly, while the membrane attack complex, formed by the terminal complement factors, is important for the killing of N. meningitidis [63]. Patients with impaired opsonization or complement activation (e.g., patients with sickle cell disease or complement deficiencies) have a higher susceptibility to invasive infections due to N. meningitidis and S. pneumoniae [64–67].

The polysaccharide capsule, although impeding adherence to the mucosal surface and entry into epithelial cells [10, 68], is the primary survival factor in the bloodstream [69]. It shields bacteria against circulating antibodies [70, 71], complement-mediated bacterial killing [72], and neutrophil phagocytosis [73]. Meningeal pathogens may also manipulate the complement cascade by enhancing the activity of regulatory proteins [74–79].

Intravascular survival of the intracellular pathogen L. monocytogenes depends mainly on its ability to evade intracellular killing in the phagolysosome [80]. Bacterial clearance by phagocytosis is an important early step in listeriosis, but the intracellular survival of L. monocytogenes facilitates evasion of innate immune mechanisms (“Trojan-horse” mechanism).

In summary, the intricate defense mechanisms of bacteria against the host immune system only need to protect few organisms, which may then cause meningitis [81].

2.1.4 Meningeal Invasion

The CNS is a tightly controlled environment. The boundary to the blood is provided by the endothelial BBB, the epithelial BCSFB, and the arachnoid [82]. Blood-borne pathogens must cross the BBB or the BCSFB to cause bacterial meningitis. Clinical observations and experimental studies link the magnitude of bacteremia with the risk of meningitis development for most meningeal pathogens [65, 83, 84]. High bacterial load in the blood alone, however, is not sufficient for the development of bacterial meningitis. The expression of selected bacterial adhesion and invasion factors is necessary for the invasion of the CNS [85].

Meningeal pathogens may migrate into the CSF by transcellular or paracellular migration [85]. The BCSFB may be more vulnerable to bacterial migration via the paracellular route because the tight junctions between the epithelial cells have a lower electrical resistance [82]. For L. monocytogenes, the previously described Trojan-horse mechanism has also been implicated in the invasion of the CNS [86–88].

Direct invasion of endothelial and epithelial cells by meningeal pathogens involves ligand–receptor interactions between bacteria and host cells followed by rearrangements of the actin cytoskeleton. This ultimately leads to the uptake of the pathogen in a vacuole and transcellular transport [42, 85, 89]. Attachment of meningeal pathogens to the tight endothelium of the BBB and the fenestrated endothelium of the BCSFB is facilitated by receptors on the apical side of the endothelial cells, some of which may be upregulated by inflammation. The 37/67-kDa laminin receptor is a shared target of cytotoxic necrotizing factor 1 (CNF1)-expressing E. coli K1, S. pneumoniae, N. meningitidis, and H. influenzae [90–94]. Bacterial adhesins mediate adhesion to the 37/67-kDa laminin receptor, including CNF1 in E. coli, choline-binding protein A (CbpA) in S. pneumoniae, secretin PilQ and porin PorA in N. meningitidis, and porin OmpP2 in H. influenzae [92, 93, 95]. S. pneumoniae may also bind to the cell adhesion molecule platelet endothelial cell adhesion molecule-1 (PECAM-1) and pIgR [96, 97].

Fimbriae and pili, hair-like structures frequently found on the surface of bacteria, are instrumental to bacterial adhesion. Type IV pili mediate attachment, bacterial aggregation, and persistence of N. meningitidis on brain endothelial cells [98]. Binding of E. coli K1 to brain endothelial cells is facilitated by FimH on type 1 fimbriae CD48 and the association of the outer membrane protein A (OmpA) with glycoproteins on the cell surface [85]. Attachment of GBS to brain endothelial cells is facilitated by several adhesins including fibrinogen receptor FbsA, laminin receptor Lmb, and the pilus tip adhesins PilA and Srr1 [99–103].

The exact mechanisms by which S. pneumoniae invades the BBB are not yet known. Partially conflicting in vitro and in vivo data indicate that binding of S. pneumoniae to transmembrane receptors pIgR and PAFr may trigger clathrin-mediated cellular endocytosis [41, 97, 104]. Specifically, the interaction of phosphorylcholine with PAFr and expression of NanA seem to be necessary for the invasion of the BBB by S. pneumoniae [40, 105–107].

E. coli invades the BBB following the induction of cytoskeletal rearrangements and actin condensations through the activation of the Rho family GTPases [85, 108, 109]. The interaction of several E. coli proteins with the BBB has been shown to facilitate cytoskeletal rearrangements culminating in the uptake of E. coli into a vacuole [94, 110–117].

For GBS, experimental data support an important role of the fibronectin-binding protein SfbA and pili in BBB invasion [101, 118]. The expression of fibronectin-binding protein OpC has also been shown to be instrumental to the invasion of the BBB by N. meningitidis [119]. N. meningitidis also form bacterial microcolonies on the luminal face of brain endothelial cells followed by the reorganization of the intracellular cytoskeleton. This may lead to the opening of intercellular junctions followed by paracellular penetration of the BBB [98]. Receptor-mediated signaling, as well as local inflammation with the recruitment of leucocytes, may further modify tight junctions and allow paracellular penetration [98, 101].

Conflicting evidence exists, whether the transcellular invasion of brain endothelial cells by L. monocytogenes may be mediated by InlA and InlB in a similar fashion as in the intestine and placenta [86, 120].

Bacteremia-independent invasion of the CNS may follow focal infections of structures close to the brain (e.g., otitis media, mastoiditis, and sinusitis) or after disruption of the integrity of the skull and meninges (e.g., malformations, trauma, neurosurgery). In a case series of 87 patients with pneumococcal meningitis, more than half of the patients had radiological signs consistent with otitis or sinusitis [121]. In a large observational study, otitis and sinusitis were reported as predisposing conditions in 25 % of patients [122]. This is supported by an experimental model of S. pneumoniae meningitis, in which a galU mutant and its parent pneumococcal strain both caused meningitis following otitis media infection in gerbils, despite the mutant’s impaired ability to disseminate to the bloodstream following infection [123].

For L. monocytogenes, S. pneumoniae and N. meningitidis retrograde access to the CNS via the neural route has also been documented [86, 124, 125]. For L. monocytogenes, however, this mechanism is likely only important in ruminants. These observations support the notion that meningeal pathogens can gain access to the CNS by several routes. The relative importance of the different routes of infection in bacterial meningitis is not exactly known.

3 Induction and Modulation of Inflammation

Our understanding of the cascade of events leading to brain inflammation during meningitis is mostly based on experimental models, in which bacteria are inoculated intranasally, but more often intracisternally or intracerebrally. These models do not fully replicate the development of meningitis as observed in patients, especially when considering the initial anatomical progression from the nasopharynx.

Early recognition of the pathogens involves brain vascular endothelial cells or perivascular macrophages [126], as well as ependymal cells. These may be aided by a trafficking population of central memory T cells which interact with local antigen-presenting cells and initiate inflammation. The very recent discovery of functional lymphatic vessels in the CNS, however, may considerably change the perception on the role of adaptive immunity in the brain [127]. Leukocytes, mainly neutrophils in the case of bacterial meningitis, are initially excluded from the CNS, being recruited later during infection [124–126]. Most pathogens responsible for bacterial meningitis have been shown to induce inflammation in brain microvascular endothelial cells in vitro, which facilitates their invasion through the blood–brain barrier. In experimental GBS meningitis, the early induction of tumor necrosis factor-α (TNF-α) and interleukin (IL)-1β mRNA expression has been documented in the ependyma and the meninges [127]. Little is known about the role of perivascular macrophages, but their depletion has been shown to worsen the clinical symptoms, to increase bacterial counts in the CSF, and to decrease leukocytes extravasation. However, since higher levels of CSF inflammation were still detected in depleted animals, a local production of inflammatory mediators in the parenchyma was also suggested [128].

During disease progression, cells in the parenchyma, especially next to the site of the initial inflammation, also express inflammatory factors, including TNF-α and IL-1β [127]. Microglia and astrocytes are sentinels in the brain parenchyma that are able to detect invading pathogens. It is postulated that, as the infection/inflammation progresses, diffusible bacterial products or host-derived factors penetrate the parenchyma, triggering a secondary response.

How the innate immune system recognizes invading pathogens is mostly determined by its ability to “sense” the bacterial surface. Bacterial pathogens causing meningitis usually are encapsulated microorganisms. The polysaccharide capsule is a virulence factor which is necessary for establishing invasive disease by preventing opsonophagocytosis (see above). In vivo, the pneumococcal capsule could also impair recognition by the innate immune system, in particular the toll-like receptor-mediated pathway [129]. The capsule can partially cover subcapsular bacterial components, depending on its thickness. For gram-positive bacteria, immunogenic components comprise teichoic (TA) and lipoteichoic acids (LTA) and peptidoglycan. For gram-negative bacteria, the endotoxin lipopolysaccharide (LPS) is recognized by the innate immune system. These components are released either by autolysis, when bacteria reach sufficient high density, or by the effect of bacteriolytic antibiotics and trigger the inflammatory reaction in bacterial meningitis. Specialized extracellular (i.e., toll-like receptors, TLRs) or intracytoplasmic (i.e., NOD-like receptors, NLRs) receptors can sense the major meningeal pathogens [130–133]. Most cells of the brain express these receptors, with microglia expressing the full repertoire of TLRs [134]. Brain endothelial cells and macrophages are also known to express these receptors. Negative outcome during bacterial meningitis has been associated with genetic variations in these pattern recognition receptors (PRR) [135–137]. TLR2 detects cell wall components of gram-positive bacteria, while TLR4 recognizes those of gram-negative bacteria. TLR9 recognizes bacterial DNA and other components. Scavenger receptors (SRs), initially described for their role in the identification and removal of modified lipoproteins, are found at the surface of perivascular and meningeal macrophages [138]. SRA-1 (CD204) and SRA-2 (MARCO) can detect N. meningitidis or S. pneumoniae [138]. The G-coupled receptor protein formyl peptide receptor-like-1 (FPRL-1), which increases the expression of antimicrobial peptide CRAMP, recognizes S. pneumoniae. Of note, it seems that the resident immune system in the brain parenchyma may already be activated during the initial presence of pneumococci in the blood in a bacteremia-derived meningitis model, even before the development of meningitis and the recruitment of neutrophils [139].

3.1 Cytokines and Chemokines

An increase in cytokines and chemokines expression levels has been documented in the brain parenchyma or the CSF of patients with bacterial meningitis and experimental models [140–143]. Cytokine and chemokine profiles vary in vitro and in experimental infection models in response to different meningeal pathogens. These differences are presumably related to pathogen-specific activation of PRR [144–147] and may in part explain the differences in mortality and morbidity observed in patients affected by the different pathogens.

The early release of TNF-α, IL-1β, and IL-6 in the course of bacterial meningitis is part of the initial innate immune answer. The expression of TNF-α and IL-1β is detected first in the ependyma and the meninges and later in the parenchyma in experimental meningitis [127]. The initial immune reaction induces a cascade of inflammatory mediators, including other cytokines and chemokines, antimicrobial peptides, prostaglandins, MMPs, NO, and ROS.

During bacterial meningitis, TNF-α activity is kept at a sustained high level in the CSF due to the continuous release of bacterial products into the CSF or by a positive feedback loop in the inflammatory cascade [148]. Similar to PPR engagement, binding of TNF-α to the TNF-α receptor leads to NF-kB activation resulting in the expression of further mediators described hereafter. Experimental administration of TNF-α into the CSF of rabbits or rats induces BBB breakdown, neutrophil influx, and increased cerebral blood flow (CBF), recapitulating pathophysiologic changes characteristic of bacterial meningitis in humans [149–154].

IL-1β is released by mononuclear phagocytes, glial cells, and endothelial cells in the CNS. Caspase-1 in the inflammasome complex activates the precursors of IL-1 or IL-18 [132, 155–157]. The CSF concentration of IL-1β significantly correlates with other inflammatory parameters and is a predictor of adverse disease outcome [158]. When experimentally injected into the CSF, IL-1 triggers a meningeal inflammation without detectable TNF-α activity, whereas the concomitant injection of TNF-α and IL-1 results in a synergistic increase in leukocyte influx into the CSF [159].

IL-6 is produced by monocytes, endothelial cells, and astrocytes, essentially in response to IL-1β and TNF-α. Its presence in the CSF of patients with bacterial meningitis is not correlated with any of the indices of meningeal inflammation or with disease severity [160]. IL-6 efficiently induces the expression of acute-phase proteins, fever, leukocytosis, and activation of the complement and clotting cascades [161]. IL-6 also possesses anti-inflammatory properties, by inhibiting TNF-α and IL-1β production in vitro and inducing IL-1 receptor antagonist [162].

Chemokine levels, including IL-8 (CXCL8), CXC5 (ENA-78), CXCL1 (GROα), and monocyte chemoattractant protein-1 (MCP-1, CCL2), MIP-1 (CCL3), and MIP-1 (CCL3), are elevated in the CSF of patients with bacterial meningitis [163, 164]. Cells shown to produce these chemokines upon stimulation include monocyte macrophages, polymorphonuclear leukocytes, endothelial cells, astrocytes, microglia, and neurons. Chemokines primarily activate and attract leukocytes to the site of inflammation. By enhancing neutrophil adhesion to endothelial cells, IL-8 regulates the migration of neutrophils toward the CNS during bacterial meningitis [142].

IL-10 is an anti-inflammatory cytokine that inhibits the production of TNF-α, IL-1, IL-6, and IL-8 in vitro. High levels of IL-10 have been found in the CSF of patients with bacterial meningitis. In experimental meningitis, application of IL-10 decreased brain edema and down-modulated subarachnoid space inflammation [165]. In addition to IL-10, transforming growth factor-β (TGF-β) also possesses anti-inflammatory properties, as demonstrated in experimental meningitis, where its application reduced cerebral edema, intracranial pressure, and CBF [166]. Similarly, IL-19 is produced by astrocytes in response to S. pneumoniae infection and acts as an anti-inflammatory modulator [167].

In bacterial meningitis, especially caused by S. pneumoniae, IFN-γ levels are increased in the CSF [157, 168]. IFN-γ stimulates nonspecific defense mechanisms such as phagocytosis or cytokine release by macrophages and polymorphonuclear leukocytes. In experimental pneumococcal meningitis in mice, the inflammasome complex initiates the production of IFN-γ [157]. IFN-γ inactivation results in increased survival, attenuation of pro-inflammatory cytokines in the CSF, less brain injury, and improved neurofunctional outcome [169].

3.2 Matrix Metalloproteinases and Related Proteins

MMPs are important role players to the pathogenesis of bacterial meningitis, participating in the breakdown of the BBB and the intrathecal production of cytokines, therefore contributing to the pleocytosis of neutrophils in the CSF [170, 171]. As endopeptidases are able to degrade ECM components, MMPs facilitate cell migration, tissue remodeling, and cytotoxicity. MMPs also modulate cytokine production and release. MMPs and related metalloproteinases, e.g., TNF-α converting enzymes (TACE/ADAM-17), transform membrane-bound cytokines, cytokine receptors, and adhesion molecules to their soluble forms by virtue of their sheddase activity. In return, cytokines such as TNF-α, IL-1, and IL-2 modulate the expression and regulation of MMPs. Most MMPs are not constitutively expressed but produced in response to increased levels of cytokines, eicosanoids, growth factors, and the presence of pathogens. Negative regulation of MMP activity is mediated by tissue inhibitors of metalloproteinases (TIMP), the specific endogenous inhibitors of MMPs. These inhibit MMPs by the formation of complexes with pro-peptide containing inactive and cleaved activated forms of MMPs. Similar to MMPs, TIMPs are also regulated in response to changing levels of a variety of signaling molecules [172].

Levels of MMP-1, MMP-3, MMP-7, MMP-8, MMP-9, and MMP-10 have been found to be elevated in the CSF of patients with bacterial meningitis. Modest increases in TIMP-1 and TIMP-2 levels have also been observed while TIMP-4 was concomitantly downregulated in the CSF of the same patient cohort [173–175]. In the brain tissue of patients with purulent meningoencephalitis, endothelial cells and infiltrating leukocytes have been shown to produce MMP-9, TIMP-1, and TIMP-2 [176].

In pediatric patients with bacterial meningitis, MMP-9 levels in the CSF fell by 90 % within 1 week, while TIMP-1 levels continued to rise by 51 % [177]. Persistent high levels of MMP-9 may be associated with a negative neurological outcome in childhood bacterial meningitis [177, 178].

In experimental pneumococcal meningitis, changes in MMP-9/TIMP-1 ratio and an increase in collagen degradation were observed, resulting in cortical brain damage [179]. MMP inhibitors prevented this effect [180, 181]. Also, combined inhibition of MMP and TACE led to a reduction of hippocampal apoptosis and preserved learning capacity of animals that recovered from bacterial meningitis [148, 182].

3.2.1 Neutrophil Invasion

Marked neutrophil pleocytosis in the CSF is a distinctive feature of bacterial meningitis. In response to chemotactic stimuli, neutrophils extravasate across the fenestrated endothelium followed by an inverse transmigration across the tight epithelial ependyma to accumulate in the CSF where they contribute to the deleterious effects of inflammation in the brain [183–185].

Inflammation induces E-selectin and P-selectin expression on the surface of endothelial cells. This allows the binding of neutrophils through P-selectin glycoprotein ligand-1 and other glycosylated ligands [186]. Deficiency in E-selectin or P-selectin expression resulted in almost complete inhibition of neutrophil influx when recombinant IL-1β and TNF-α were directly injected into the CSF of mice [187]. Blocking of selectins by the polysaccharide fucoidan also attenuates CSF neutrophil pleocytosis in experimental pneumococcal meningitis [188–190]. The firm adhesion of chemokine-activated leukocytes to the endothelium is mediated by the binding of integrins to members of the immunoglobulin-like superfamily on the endothelium, including ICAM-1 and vascular cell adhesion molecule-1 (VCAM-1). Antibody-mediated blocking of integrins or ICAM-1 decreases leukocyte influx in experimental meningitis models, confirming an important role for these molecules in assisting leukocyte entry into the inflamed neural tissue [191–193].

Bacteria-induced transmigration of leukocytes across the BBB or the BCSFB may occur by the paracellular or transcellular route as determined in in vitro models [194–197]. Intracellular signaling induced by the interaction of integrins on the leukocytes with endothelial cell adhesion molecules leads to structural changes in endothelial cells including remodeling of the actin cytoskeleton. This facilitates migration of leukocytes either transcellularly or paracellularly [198]. Interestingly, the early presence of leukocytes in the CSF does not affect bacterial multiplication in experimental meningitis models [199, 200].

3.2.2 Resolution of the Inflammation

Apart from the production of anti-inflammatory activity of IL-6, IL-10, and TGF-β, different mechanisms are initiated later during the disease course and participate in the resolution of inflammation. The clearance of invading neutrophils by apoptosis is necessary to avoid the excess release of neurotoxic molecules. Experimental inhibition of apoptosis in neutrophils by overexpression of the antiapoptotic factor bcl-2 prolonged the expression of pro-inflammatory cytokines in experimental pneumococcal meningitis in mice. On the contrary, treatment with roscovitine, a specific inducer of apoptosis in neutrophils, reduced damage and improved survival in vivo in the same model [201]. Similarly, TNF-related apoptosis-inducing ligand (TRAIL) released by microglia triggers leukocyte apoptosis. Elevated levels of TRAIL are documented in the CSF of patients with pneumococcal meningitis. The administration of recombinant TRAIL increased leukocyte apoptosis and decreased inflammation and neuronal apoptosis in experimental pneumococcal meningitis, while TRAIL knockout animals displayed a worse outcome than wild-type mice [202].

4 Sites of Damage in Meningitis

Derived from observations in patients and experimental models of bacterial meningitis, the cerebral vasculature, the brain parenchyma (cortex or hippocampus), and the inner ear are primarily injured by the disease. The morbidity observed in survivors reflects damage to these structures.

Figure 3 shows the histopathology of experimental pneumococcal meningitis and group B streptococcal meningitis.

Fig. 3
figure 3

Histopathology of experimental pneumococcal meningitis (a–c, e) or group B streptococcal meningitis (d). (a) Extensive cortical injury consisting of areas of cortical necrosis. A wedge-shaped distribution of reduced neuronal density, suggestive of ischemic damage, can be distinguished from neighboring more healthy cortical tissue (cresyl violet, magnification × 10). (b) Cortical neuronal loss on the right contains necrotic neurons characterized by cell swelling and fading of cytoarchitecture and is sharply demarcated from preserved neurons on the upper left (cresyl violet, magnification × 40). (c) Hippocampal dentate gyrus histology of an infant rat suffering from pneumococcal meningitis at 42 h after infection. Apoptotic cells are characterized by the presence of round or oval apoptotic bodies consisting of regularly shaped, often fragmented, dark chromatin clumps (upper left inset). Apoptotic cells are mostly observed in the inner rim (i.e., the subgranular zone) of the dentate gyrus (cresyl violet, magnification × 40). (d) Dentate gyrus of the hippocampus of an infant rat infected with group B streptococcus. Large clusters of cells with uniformly dense and shrunken (pyknotic) nuclear morphology (upper left inset) are observed throughout the entire blade. (e) Subarachnoid space inflammation consisting of bacteria and inflammatory cells (asterisk) extending into the Virchow-Robin space (arrowheads) around the penetrating cortical vasculature (cresyl violet, magnification × 63)

Full size image

4.1 Cerebral Vasculature

The cerebral vasculature is one of the primary sites involved in the development of meningitis (Fig. 3e). The early activation of endothelial cells by inflammatory mediators, the disruption of endothelial function by meningeal pathogens, and the neutrophil influx into the CSF lead to several pathological changes. These include the disruption of the BBB, brain edema, loss of CBF autoregulation, and focal and global changes of CBF resulting in cerebral ischemia. Extensive cerebral infarction, resulting from vasculitis and coagulation disturbances, and circulatory failure due to septic shock may lead to acute death in bacterial meningitis [203].

4.1.1 Pathology

Subarachnoid space inflammation is a characteristic feature of acute bacterial meningitis. Arterial narrowing has been shown to be a predominant finding in patients with arterial complication. Vasculitis or vasospasm may lead to brain ischemia. Additionally, venous thrombosis has been observed in adult patients with pneumococcal meningitis [121, 204, 205]. In fatal cases of neonatal meningitis, inflammatory vasculitis is uniformly present, indicating that the cerebral vasculature of the neonate may be particularly susceptible to inflammatory damage, leading to notably severe structural damage to the neonatal brain [206].

4.1.2 Disruption of the Blood–Brain Barrier

In meningitis, the permeability of the BBB and BCSFB increases as the result of functionally relevant alterations induced by the disease process. Mainly paracellular leakage has been studied in this respect. The remodeling of the actin cytoskeleton, the reorganization of tight junctions, and the enzymatic degradation of tight junctions and basement membrane components participate in the permeabilization of the BBB [198, 207].

The meningeal pathogens are able to induce direct damage to endothelial or epithelial cells. In vitro cytotoxic damage to brain endothelial cells has been seen with S. pneumoniae, mainly mediated by pneumolysin [208–210], N. meningitidis, primarily mediated by NO [211], and E. coli K1 [212]. Pneumolysin was also shown to lead to astrocyte cytoskeleton remodeling in vitro [213], which might result in BBB disruption. In experimental GBS and S. pneumoniae meningitis, increased expression of the transcriptional repressor Snail1 has been found, leading to downregulation of tight junction components, like ZO-1, claudin-5, and occludin [214].

Signaling and effector molecules released during the inflammation process and changes caused by the interaction of inflammatory cells with the brain barriers also alter the function of cells constituting the BBB. A variety of cytokines and chemokines have been shown to induce tight junction and cytoskeleton rearrangements leading to BBB disruption [215–218].

MMPs increase BBB permeability by proteolytic breakdown of the basement membrane and tight junctions. In experimental meningococcal meningitis in rats, disruption of the BBB, increased intracranial pressure, and CSF pleocytosis were observed in parallel to an increase of MMP-9 activity in the CSF 6 h after infection [170]. Experimental inhibition of MMPs reduces occurrence of intracerebral bleeding and BBB breakdown in mice with meningococcal meningitis [219].

ROS and NO also contribute to BBB disruption via lipid peroxidation and oxidative damage to cell membrane proteins [220]. Neutralization of ROS and NO by radical scavengers in experimental meningitis prevented cerebral damage including BBB breakdown [221–224]. Additionally, ROS and NO lead to tight junction and cytoskeletal reorganization and MMP activation [225].

As a consequence of the increased BBB/BCSFB permeability, potentially harmful molecules found at high concentrations in the blood can accumulate in the CSF, unfavorably modifying the brain’s microenvironment. On the other hand, CSF penetration of antibiotics is facilitated by BBB opening, therefore improving the efficiency of therapy by increasing antibiotic CSF/serum ratio in comparison to noninflammatory conditions [226].

4.1.3 Alterations of Cerebral Blood Flow

Marked changes in CBF are observed during bacterial meningitis. Brain edema, obstructive hydrocephalus, meningitis-associated cerebritis, cerebral infarction and cerebral venous thrombosis, and status epilepticus may all lead to intracerebral hypertension [227, 228]. In the advanced disease stages, CBF is usually reduced [204, 229, 230]. Focal changes in the vasculature, loss of CBF autoregulation, intracranial pressure, and systemic hypotension perturb CBF in bacterial meningitis.

In humans, cerebrovascular changes in bacterial meningitis include segmental narrowing of vessels, irregularities of vessel walls, and arterial and venous thromboses [121, 231–233]. The middle and anterior cerebral arteries or the basilar artery are frequently affected by stroke associated with bacterial meningitis. Vessel narrowing results in an increased velocity of cerebral blood, an early predictor of cerebrovascular complications [121, 204]. Sequelae of vascular complications may result in hemiparesis or quadriparesis [206, 234, 235].

Disturbance of CBF autoregulation, combined with systemic hypotension, may lead to cerebral ischemia [236–240]. Alternatively, systemic hypertension may augment vasogenic edema and intracranial pressure [240, 241], ultimately limiting CBF. In addition to global cerebral hypoperfusion, vasculitis of large and small arteries traversing the inflamed subarachnoid space leads to regional hypoperfusion. This form of ischemic damage (Fig. 3a, b) may be responsible for permanent neurologic sequelae following bacterial meningitis [235, 236].

NO plays a crucial albeit complex role in modulating CBF during meningitis. Early in the disease, the vasodilatory effect of NO contributes to the hyperemia induced by the subarachnoid space inflammation. Later, NO generated in the vasculature provides some protection against ischemia by attenuating the effects of the progressive decline in CBF due to the production of vasoconstrictive factors (see later discussion).

ROS play a critical role in modulating CBF during meningitis [221, 242–244]. Generation of ROS is localized primarily to the cells constituting the subarachnoid and ventricular inflammation and the cerebral vasculature, as shown in infant rats with experimental meningitis [221]. In this model, the cerebral vasculature showed evidence of marked oxidative alterations, whereas oxidative damage to the brain parenchyma itself was not documented conclusively [245]. Antioxidant treatment prevents oxidative vascular damage, hyperemia, CBF decline, and rise of ICP in experimental pneumococcal and GBS meningitis [221, 222, 243, 244].

The reaction of NO with ROS produces peroxynitrite, a strong oxidant that exerts cytotoxic effects [220, 246]. Nitrotyrosine residues, a reaction product of peroxynitrite with proteins, were detected in the meninges, the cortical blood vessels, and inflammatory cells in the brains of patients with bacterial meningitis and corresponding animal models [220, 247]. Treatment with urate, a peroxynitrite scavenger, reduced the meningeal inflammation, BBB disruption, and intracranial hypertension [247].

Increased endothelin levels are found in the CSF of patients with bacterial meningitis [248]. This potent vasoconstrictive peptide is produced in the CNS by vascular endothelial cells, astrocytes, and neurons and participates in CBF regulation. Endothelin synthesis is triggered by cytokines, i.e., TNF-α [249–251], and inhibited by NO [252]. In experimental pneumococcal meningitis, an endothelin antagonist (bosentan) significantly prevented the reduction of CBF and attenuated the extent of cerebral ischemia [253].

4.2 Inner Ear

Unilateral or bilateral sensorineural hearing loss (SNHL) is the most common neurologic sequelae following bacterial meningitis and is found in 5–30 % of survivors [254–257], with S. pneumoniae causing the highest rate of sensorineural hearing loss [258–261]. Hearing loss during bacterial meningitis is progressive rather than abrupt, and its magnitude depends on the duration of untreated infection [255]. Hearing loss is the direct result of inflammation in the inner ear during the acute phase of the disease. Magnetic resonance imaging (MRI) studies in humans with meningitis have confirmed the inflammatory involvement of the inner ear [262]. In survivors of meningitis, progressive cochlear ossification and spiral ganglion loss are observed years after the disease [263, 264].

4.2.1 Pathology

During the acute stage of meningitis, suppurative labyrinthitis is usually observed in the human temporal bone of patients, often accompanied by purulent infiltrate in the perilymphatic duct [265]. Studies in experimental meningitis have shown that bacteria and inflammatory cells are present in the cochlea in the earliest stages of pneumococcal infection [266, 267]. Inflammatory infiltration of the cochlea progresses via the cochlear aqueduct to the perilymphatic space (scala tympani) and via the spiral ligament to the endolymphatic space [267–269]. Damage to the blood–labyrinth barrier, the hair cells, and the spiral ganglion is observed first at the base and later at the apex of the cochlea, corresponding to hearing loss first at high then at low frequencies [270]. The first signs of damage and hearing loss are observed for hair cells with a partial reversibility of associated hearing loss [270]. Later during the course of the disease, irreparable ultrastructural inner ear damage, especially the loss of spiral ganglion neuronal cells, is associated with severe to profound deafness [266, 271]. Toxic effects of the meningeal pathogen (e.g., pneumolysin from S. pneumoniae) and inflammatory mediators appear to be responsible for the cytopathic effects [272, 273]. In particular, the production of ROS and NO has been involved in the pathogenesis of cochlear damage and hearing loss in bacterial meningitis [270]. Treatment with peroxynitrite scavengers or antioxidants attenuated hearing loss and protected spiral ganglion neuronal cells [274, 275].

4.3 Central Nervous Tissue

4.3.1 Neurologic Sequelae

Bacterial meningitis causes damage to both cortical and subcortical brain structures. The neurologic sequelae resulting from brain damage include hearing impairment (see previous discussion), obstructive hydrocephalus, focal sensory-motor deficits, mental retardation, seizure disorders, and cortical blindness. Behavioral and cognitive sequelae in children and adults after bacterial meningitis are common [276–281]. Morbidity in patients surviving bacterial meningitis is usually higher with S. pneumoniae as causative pathogen [122, 254, 282].

Follow-up studies revealed that neurologic sequelae of childhood meningitis persist for more than 10 years, impacting the school performance of affected children [277, 279, 283, 284]. In adults tested 0.5–13.5 years after acute bacterial meningitis, 32 % of 155 meningitis survivors suffered from relevant impairment of psychomotor performance, speed of cognitive processes, and concentration and memory functions [280].

4.3.2 Pathology

Histological changes are often seen in a focal pattern (Fig. 3a, b). Infarctions in the frontal, temporal, and parietal lobes, the basal ganglia, the thalamus, as well as periventricular white matter injuries are found in newborns or children affected by bacterial meningitis, leading to cerebral atrophy and hydrocephalus [205, 285–287]. Structural changes were also seen in the temporal lobe and the limbic system of adult patients [281]. Neuronal loss is associated with a marked activation of astrocytes and microglia [288], as well as axonal injury [289, 290]. Furthermore, meningitis induced by gram-negative bacteria is often characterized by the development of brain abscesses [291, 292].

Apoptotic cell death has been observed in the dentate gyrus of the hippocampus during meningitis [221, 293, 294] (Fig. 3c) with the presence of condensed, fragmented nuclei (Fig. 3c, inset), and the detection of fragmented DNA and active caspase-3, an effector enzyme involved in the execution of programmed cell death [295]. A second form of hippocampal neuronal damage with morphologically distinct features (uniformly shrunken nuclei, clusters of damaged cells) may also be found in the lower blade of the dentate gyrus cell, spanning the entire width of the dentate gyrus band (Fig. 3d). This form of hippocampal damage is reminiscent of ischemia-related neuronal damage [296] and is the preferential pattern of neuronal injury observed in meningitis caused by GBS [216, 294].

The neuronal injury in the hippocampus is of particular significance because experimental data suggest that it is related to learning impairment following meningitis [148, 297, 298]. The observation of cell death primarily of immature progenitor cells in the subgranular zone of the dentate gyrus, cells necessary for the acquisition of new memory, strengthens this hypothesis [294, 295, 299]. Cognitive impairment and learning disabilities following meningitis may thus be reflected by damage to the dentate gyrus of the hippocampus [148, 297, 298, 300].

Confirming the findings in animal models, brain sections of patients who died from bacterial meningitis showed apoptotic neurons with cells staining for the active form of caspase-3 in the dentate gyrus [301]. Also, volumetric measurements of the hippocampus by MRI techniques showed unilateral and bilateral hippocampal atrophy in patients surviving meningitis [302], potentially reflecting the apoptotic loss of neurons observed by histopathology.

4.3.3 Brain Edema and Cerebral Herniation

Cerebral edema in bacterial meningitis may be the combined result of vasogenic, cytotoxic, interstitial, or osmotic edema [296]. Vasogenic cerebral edema is the consequence of increased BBB permeability, with the extravasation of plasma proteins into the brain parenchyma (see earlier discussion) and mainly affects the white matter [296]. Cytotoxic edema is an increase in intracellular water due to the intracellular accumulation of osmotically effective ions (i.e., sodium, potassium, or glutamate). Cytotoxic mechanisms include ischemia and the effect of excitatory amino acids (EAA) [296, 303, 304]. Cytotoxic edema affects the gray and white matter. The increase of CSF influx due to a more sustained production (increased blood flow in the choroid plexus) or a decreased resorption (increased CSF outflow resistance across the arachnoid villi system of the sagittal sinus) is the cause of interstitial edema [305]. Finally, osmotic edema may be caused by inappropriate secretion of antidiuretic hormone during the course of bacterial meningitis. The observed hypoosmolality of serum results in a net influx of water into the brain [296, 306].

Brain edema is an important contributor to the fatal outcome of bacterial meningitis [121, 307, 308]. Extensive brain edema leads to increased intracranial pressure culminating in the herniation of brain tissue and compression of the brainstem.

Aquaporins (AQP) are pore-forming membrane proteins regulating water homeostasis of the brain [309]. In cytotoxic edema, the transcellular influx of water across the BBB is facilitated by AQP-4, while in vasogenic and interstitial brain edema, the absence of AQP-4 leads to more severe brain edema by limiting outflow of CSF [296, 310]. In a mouse model of pneumococcal meningitis, AQP-4 was significantly upregulated and associated with increased cytotoxic brain edema [311]. AQP-4 upregulation was also found in the brain of a patient with bacterial meningitis [312], but not in the CSF of patients with bacterial meningitis [313].

4.3.4 Mediators of Cell Death in Neuronal Tissue

EAAs including glutamate induce neuronal apoptosis and necrosis and appear to mediate cellular injury in a variety of brain disorders. In bacterial meningitis, ischemia and EAA may cause direct neuronal toxicity as suggested from experimental models or clinical studies [288, 304]. In patients and experimental pneumococcal meningitis in rabbits, CSF glutamate concentrations were significantly elevated and correlated with the severity of the disease [314, 315]. The cause for increased concentrations of EAA in the brain during meningitis is still poorly understood, but may be related to BBB disruption. Alternatively, pore-forming cytolysins, like pneumolysin, may cause the release of glutamate from astrocytes, leading to subsequent glutamate-dependent synaptic damage [316]. Kynurenic acid, a nonselective inhibitor of the neurotoxic effect of EAA acting on NMDA receptors, significantly attenuated brain injury, both in the cortex and in the hippocampus in an infant rat model of neonatal meningitis [288, 304].

The intensity and duration of the inflammatory response in bacterial meningitis correlates with the development of neuronal damage. Meningeal pathogens interfere with the programmed cell death of immune cells, possibly prolonging inflammation. For example, N. meningitidis inhibits apoptosis in neutrophils [317] or macrophages [318]. Similarly GBS interfere with TLR-2-mediated programmed cell death in macrophages and microglia [319].

The brain is one of the most metabolically active organ in the body, therefore requiring a high rate of oxygen consumption. But because it possesses a high level of polyunsaturated fatty acid prone to lipid peroxidation, the brain is particularly vulnerable to oxidative damage caused by ROS. This is also exacerbated by the high abundance of redox-active metals (iron or copper) and the relative low levels of endogenous antioxidant [320]. The high concentration of mitochondria in cerebrovascular endothelial cells might also account for the sensitivity of the BBB to oxidant stressors [321]. The prevention of BBB disruption, attenuation of lipid peroxidation, reduction of ischemia in the brain parenchyma, and the reduced neuronal injury observed with the use of radical scavengers in experimental meningitis underline the role of ROS [322]. Despite these observations, a direct neurotoxic effect of ROS in bacterial meningitis has not been documented conclusively. Most of the beneficial effects of ROS scavengers in experimental meningitis may be primarily due to a beneficial effect at the level of the cerebral vasculature [245] (see previous discussion).

NO may be either neuroprotective or neurotoxic in bacterial meningitis. In its oxidized form, it can inactivate glutamate receptors, therefore reducing EAA-mediated excitotoxicity. In its reduced form, it reacts with superoxide and forms peroxynitrite, which is highly reactive and causes damage to DNA, lipid membranes, and proteins [323]. Caspase-3-dependent hippocampal apoptosis was reduced during experimental pneumococcal meningitis in mice lacking iNOS expression [324]. In contrast, iNOS inhibition by aminoguanidine exacerbates neuronal injury in experimental GBS meningitis [325].

MMPs may exert direct neurotoxic effects by degrading perineuronal components of the ECM, like laminin [326, 327], or the neural cell adhesion molecule NCAM [328]. Indirectly, the shedding of death receptors ligands (FasL, TNF-α) by MMPs can induce death in cells that express these ligands, including neurons [329]. Given as adjuvant therapy, MMP inhibitors reduced the extent of cortical damage, and combined inhibition of MMP and TACE led to a reduction in hippocampal apoptosis [148, 180, 182]. MMP-TACE inhibitors also improved learning capacity after experimental pneumococcal meningitis in infant rats [148].

Bacteria themselves may also directly damage neurons. Pneumococci have been shown to induce cell death in microglia or neurons in vitro [330]. Pneumolysin and superoxide produced by S. pneumoniae have been identified as the responsible neurotoxic mediators. In vitro, these toxins cause neuronal death via damage to mitochondria with the subsequent release of the proapoptotic mitochondrial factors cytochrome c and AIF [331, 332]. In experimental pneumococcal meningitis in rabbits, pneumolysin co-localized with apoptotic neurons of the hippocampus, and infection with pneumococcal mutants deficient for pneumolysin and superoxide production caused significantly less damage [331]. Pneumolysin was also shown to induce cochlear hair cell death in the rat [333] and endothelial cell damage in vitro [209, 210]. Other bacterial pore-forming toxins such as the beta hemolysin/cytolysin of GBS [334] are able to directly induce neuronal damage.

5 Conclusion

The pathophysiology of bacterial meningitis involves a complex interplay between the invasive pathogen, the anatomical barriers of the CNS, and the immune defense system of the host. In that context, the BCSFB and BBB play a primordial role, by restricting access to the brain. Meningeal pathogens have evolved sophisticated strategies to cross the otherwise hard-to-reach CNS compartment. Doing so, the successfully invading pathogens initiate a catastrophic cascade of events, involving the permeabilization of the barrier membranes, the massive recruitment of neutrophils, and a deleterious inflammatory reaction. Protecting barrier integrity by targeting pathophysiological processes that lead to its impairment, as demonstrated from experimental evidence using pharmacologic interventions including, e.g., radical scavengers or MMP inhibitors, is a promising strategy with the potential to reduce neurofunctional impairment in survivors of the disease.