The role of pneumolysin in pneumococcal pneumonia and meningitis

Abstract

Diseases caused by Streptococcus pneumoniae include pneumonia, septicaemia and meningitis. All these are associated with high morbidity and mortality. The pneumococcus can colonize the nasopharynx, and this can be a prelude to bronchopneumonia and invasion of the vasculature space. Proliferation in the blood can result in a breach of the blood–brain barrier and entry into the cerebrospinal fluid (CSF) where the bacteria cause inflammation of the meningeal membranes resulting in meningitis. The infected host may develop septicaemia and/or meningitis secondary to bronchopneumonia. Also septicaemia is a common precursor of meningitis. The mechanisms surrounding the sequence of infection are unknown, but will be dependent on the properties of both the host and bacterium. Treatment of these diseases with antibiotics leads to clearance of the bacteria from the infected tissues, but the bacteriolytic nature of antibiotics leads to an acute release of bacterial toxins and thus after antibiotic therapy the patients can be left with organ-specific deficits. One of the main toxins released from pneumococci is the membrane pore forming toxin pneumolysin. Here we review the extensive studies on the role of pneumolysin in the pathogenesis of pneumococcal diseases.

INTRODUCTION

The pneumococcus is an important human pathogen that colonizes the upper respiratory tract. This can lead to diseases of high morbidity and mortality such as pneumonia, septicaemia and meningitis. A key virulence factor in these events is the pneumococcal toxin, pneumolysin, which is a 53Kd protein produced by virtually all clinical isolates of the pneumococcus [1]. Pneumolysin is classically defined as a pore-forming toxin that is inhibited by cholesterol [2,3]. It is common to all serotypes of S. pneumoniae, and it can be thought of as a multi-effective factor for virulence following pneumococcal infection. At high (above 50 haemolytic units) levels it is lytic to all cells with cholesterol in the membrane. The lytic activity of this toxin can be inhibited by preincubation with cholesterol, consistent with the suggestion that membrane cholesterol is the receptor for this toxin [4]. At lower, sublytic concentrations, which probably exist in the early stages of infection, the toxin also may cause a range of effects, including induction of apoptosis [5], activation of host complement [6] and induce proinflammatory reactions in immune cells [7]. At higher lytic concentrations, which may exist in the later stage of infection, the toxin may cause widespread direct cellular and tissue damage by virtue of its membrane pore forming properties [3].

An important characteristic of the pneumococcus that is thought to be essential for pneumolysin release and its subsequent virulence is the property of undergoing autolysis, which is characterized by cell wall degradation by a peptidoglycan hydrolase (autolysin). The main autolysin in the pneumococcus is N-acetyl-muramoyl-1-alanine amidase, commonly known as Lyt A [8]. It is thought that when Lyt A is activated the pneumococcal virulence factors are released. Pneumolysin is present within the bacterial cytoplasm and because it does not have a N-terminal secretion signal sequence, lysis of the bacteria is essential for its release [9].

Using isogenic mutant pneumococci deficient in pneumolysin it has been shown that these bacteria may be cleared from the lungs following infection, which provides further evidence that pneumolysin is essential for virulence [10].

Pneumolysin has been studied in a wide range of model systems in order to define its role in the pathogenesis of disease. These models include those analysing the interactions of purified pneumolysin or whole pneumococci with isolated cells or tissues in vitro[11–14]. A summary of pneumolysin activities derived from these in vitro models is listed in Table 1.

Table 1.

Examples of in-vitro biological properties of purified pneumolysin

Activity	Reference
Inhibition of polymorphonuclear cell respiratory burst, random migration and chemotaxis	[66]
Inhibition of mitogen-induced proliferation and antibody production by human lymphocytes	[67]
Activation of classical complement pathway	[4]
Lysis of erythrocytes	[68]
Inhibition of ciliary beat of respiratory mucosa	[69]
Toxic to pulmonary alveolar epithelial cells	[14]
Stimulation of TNF-α and IL-1β production from human monocytes	[31]
Activation of phospholipase A2 in pulmonary endothelial cells	[70]
Separation of epithelial cell tight junctions	[71]
Initiates nitric oxide production from macrophages	[72]
Reduces ciliary beat frequency of cerebral ependymal cells	[11]
Induces production of IFN-γ in spleen cells	[23]
Induces synthesis and release of IL-8 from neutrophils	[45]

The main focus of this review will be on studies of the role of pneumolysin in pneumococcal infection in vivo, and pneumococcal interactions with the host immune system.

THE ROLE OF PNEUMOLYSIN IN PNEUMONIA AND BACTERAEMIA

Although the existence of the pneumococcus has been known for over 100 years, it was not until the 1940s that a pneumococcal haemolysin was identified [15,16]. The physical properties of pneumolysin were described a few decades later [17], and an important role for pneumolysin in the virulence of S. pneumoniae in vivo was demonstrated conclusively in the 1980s [18]. Paton and coworkers showed that immunization of mice with a partially inactivated form of pneumolysin could produce a moderate protection from subsequent challenge with virulent pneumococci [18]. Later, following cloning and sequencing of the pneumolysin gene [19], genetically modified toxoids were made and these provided impressive protection against disease from a range of pneumococcal serotypes [20]. The cloning and sequencing of the pneumolysin gene [19] also allowed the construction of isogenic pneumolysin-negative mutants of type 2 and type 3 pneumococci, which significantly advanced our understanding of the role of this toxin in disease [10,21]. Access to the cloned pneumolysin gene enabled structure–function data to be collected and these showed that the anticellular activities of the toxin were derived from different parts of the molecule than the complement-activating activity [4]. This, in turn, has allowed further analysis of the role of pneumolysin by construction of other isogenic pneumococcal mutants expressing a version of the toxin in which one or other of these activities were altered by modification of the toxin gene. Such mutations have resulted in the production of different types of pneumolysin deficient in haemolytic activity [22,23] and/or complement activation [24]. Using isogenic mutant strains, we have examined the effect of deletion of the cytotoxic activity or complement-activating activity of pneumolysin on bacterial growth in lungs, blood and on histological changes in infected lung tissue and the pattern of inflammatory cell recruitment. Both activities of pneumolysin contributed to the pathology in the lung as well as the timing of the onset of bacteraemia [25]. Histological changes in the lungs were delayed after infection with either mutant compared to changes seen after infection with the wild-type pneumococcus. The complement-activating activity of pneumolysin significantly reduced the accumulation of T cells, whereas the toxins cytolytic activity significantly increased the neutrophil recruitment into the lung tissue.

In the early defining studies, the virulence of a type 2 pneumococcus lacking pneumolysin (designated PLN-A) was reduced 100-fold when administered intranasally to mice [10]. Also when given intravenously to mice, this pneumolysin-negative mutant survived significantly less well when compared to its wild-type parent [10]. Further studies showed that following intranasal challenge of mice with PLN-A, a significantly less severe inflammatory response, a reduced rate of bacterial multiplication within the lungs and a delayed onset of bacteraemia were observed compared to infection with the wild-type parent [26]. In a model of lobar pneumonia, resulting from intratracheal inoculation of mice, PLN-A was shown to be 10 times less virulent than its wild-type parent [27]. PLN-A exhibited a significantly decreased ability to grow in the lung and had little effect on the alveolar–capillary barrier, which resulted in reduced ability to penetrate from the alveoli into the interstitium of the lung [27].

In a model of bacteraemia, the growth rate of PLN-A was enhanced by co-infection with wild-type pneumococci [28]. This suggests that pneumolysin increases the ability of pneumococci to multiply. Benton and colleagues [28] also showed that intravenous challenge of mice with PLN-A resulted in chronic bacteraemia, with numbers of pneumococci remaining constant at around 10⁷ colony-forming units (CFU) per ml blood for several days post-infection. In contrast, the wild-type parent organism exhibited exponential growth, reaching 10¹⁰ CFU/ml in the blood by the time the animals died within 28 h of infection [28]. The authors concluded that during the first few hours of bacteraemia, pneumolysin played a crucial role by preventing the initiation of host immune responses, thereby allowing the exponential growth of pneumococci [28].

COLONIZATION AND CYTOKINE RESPONSES TO PNEUMOLYSIN

Colonization of the nasopharynx is a common prelude to invasive pneumococcal disease. We have shown that pneumolysin was essential for successful colonization of the nasopharynx [29]. Pneumolysin-sufficient wild-type pneumococci were able to colonize successfully and infect the nasopharynx, whereas pneumolysin-deficient mutant PLN-A was less able to colonize the nasopharynx and in some cases could be completely cleared [29]. The importance of pneumolysin for colonization is consistent with the in-vitro data of Rubins et al. [30], who found that pneumolysin-deficient pneumococci attached significantly less well to respiratory epithelial cells. However, a strain effect was apparent from their data, since a pneumolysin-deficient serotype 3 strain, was not significantly different from the wild-type, further indication that the contribution of the toxin can vary from strain to strain comes from a comparison of our in-vivo data. Rubins and colleagues [30] concluded that pneumolysin is not a major determinant of colonization of the murine nasopharynx on the basis of their work with a pneumolysin-deficient serotype 14 strain. Clearly, our findings [29] challenge this uncomplicated conclusion and suggest that the context of other pneumococcal factors determines the overall contribution of pneumolysin. The conclusion that the overall context of the strain determines the importance of individual factors is supported by our investigation of the role of capsule in colonization. Differences in pneumococcal serotype also altered the contribution of pneumolysin to upper and lower respiratory tract colonization [29]. In this study, we showed that differences in pneumococcal capsule type had significant effects on pneumococcal colonization and subsequent infection of nasopharynx, trachea and lungs. However, it was the combination of capsule type and genetic background that was important and the influence of this combination varied with the site of infection. For example, in the nasopharynx the wild-type serotype 3 and the capsule-switched mutant (capsule switched type 2 to type 3) behaved similarly, whereas in the lungs the mutant that was capsule-switched showed lower survival than the wild-type serotype 3 [29]. Therefore, the combination of capsule type and genetic background of the bacteria also determined the overall virulence. Thus the wild-type serotype 3 strain was virulent, whereas the capsule-switched mutant was significantly less virulent [29].

Broadly, cytokines are considered to be the key regulators of the host immune response to infection. Cytokines are proteins produced by leucocytes and epithelial cells to encourage recruitment and activation of host immune cells within the infected tissues. Evidence of cytokine responses to pneumolysin have been obtained from intravenous infection of mice with PLN-A and wild-type pneumococci. For example, intravenous infection of mice with either PLN-A or wild-type pneumococci increased plasma levels of interleukin (IL)-6 [28]. However, the plasma levels of IL6 from PLN-A infected mice were reduced compared with mice infected with wild-type pneumococci. The precise mechanism by which pneumolysin regulates IL-6 levels remains to be determined.

Pneumolysin has been shown to mediate the production of the proinflammatory cytokine tumour necrosis factor (TNF)-α[31]. Benton and colleagues showed that the host resistance to PLN-A developed soon after intravenous infection, and this was dependent on TNF-α activation. The study also reported that neither IL-1β or IL-6 had a role in the resistance of mice to PLN-A infection [32]. A series of studies suggested that following intranasal infection with wild-type pneumococci, TNF-α, IL-6 and IL-10 levels in the lung significantly increased. However, IL-6 was the only cytokine that also exhibited high levels in blood [33,34]. In addition to demonstrating increased IL-6 levels in blood, the authors also showed that mice deficient in IL-6 exhibited increased levels of TNF-α, IL-1β and interferon (IFN)-γ, in their lungs, as well as the anti-inflammatory cytokine IL-10 following pneumococcal infection. The infected animals also had significantly higher numbers of pneumococci in lungs when compared to controls; this was accompanied by shorter survival rates [34]. IL-6 is known to be secreted by alveolar macrophages in lung tissue and was found roughly at equivalent times to neutrophil infiltration during the course of infection. Its release starts to increase from 6 to 12 h post-infection onwards, peaking at 24–36 h, using mice after intranasal infection [35]. The precise role of IL-6 in the sequence of events during infection remains to be elucidated fully. Other cytokines are involved, such as IL-8. This chemokine or its equivalent in mice known as MIP-2 is released from broncho-epithelial cells and alveolar macrophages upon infection or contact with pneumococci in order to attract neutrophils to sites of pneumococcal infection. MIP-2 is released from 3 h onwards and peaks at 12 h post-infection in lung tissue (unpublished observations).

Other studies using TNF-α p55 receptor knock-out mice have shown that a TNF response is necessary to control bacterial loads within the lungs and bloodstream during pneumococcal pneumonia [35]. These findings are consistent with an earlier study, which showed increased susceptibility of p55 knock-out mice to systemic pneumococcal infections [36]. An interesting new observation was that the highly susceptible CBA/Ca murine strain had reduced levels of TNF-α in lung airways during the early stages of infection compared to the resistant strain Balb/C [37]. It may be hypothesized that low levels of TNF-α leads to reduced numbers of airway neutrophils, allowing pneumococcal numbers to increase within the lung tissue [35].

EFFECT OF PNEUMOLYSIN ON IMMUNE CELLS IN PNEUMONIA

To date, the work illustrating the requirement for pneumolysin in pneumonia is derived mainly from the pneumolysin negative strain of pneumococcus, PLN-A. The in vivo studies, focusing on the host immune response during pneumococcal pneumonia, have allowed us to understand some of the host immune cell–pneumococcal interactions [37]. Resident alveolar macrophages are the first cells that are likely to combat pneumococci in the early stages of infection of the lung. However, it is the neutrophil that is the major immune cell responsible for pneumococcal clearance from the lungs. Following pneumococcal invasion, neutrophils are recruited rapidly to sites of infection in great numbers. Indeed, the overwhelming majority of infiltrating inflammatory cells are neutrophils. For more information on the role of neutrophils in animal models of pneumococcal diseases the reader is directed to the following articles [38–44].

We know that during pneumococcal pneumonia a sequential infiltration pattern of inflammatory cells into lung tissue is observed, but in the susceptible host this fails to eliminate wild-type pneumococci from the lungs [44]. The inflammatory cell accumulation occurs gradually, with the timing of the accumulation influenced by the presence or absence of pneumolysin (Figure 1). In the absence of pneumolysin, the inflammatory cell influx is significantly delayed and is less intense [44]. The presence of pneumolysin increases the rate of influx of neutrophils. This may be caused by pneumolysin increasing the rate of local tissue damage, and thus increasing the rate of cytokine production. Alternatively, it could result from pneumolysin activation of complement [27] or neutrophil chemotactic factors from neutrophils themselves [45]. It is also possible that pneumolysin may inhibit the activity of the neutrophils on arrival to infected lung tissue by virtue of its cytotoxic activity [44]. Indeed, this hypothesis is supported partially by an earlier study of pneumolysin in vitro, whereby pneumolysin significantly depressed a variety of phagocytic functions of immune cells [46].

Fig. 1.

Schematic representation of events during the early (right side) and late stage (left side) responses, based on our mouse (MF1) intransal infection model. In the early stages of infection pneumococci infect bronchiolar airspaces, and subsequently invade the lung epithelial cells. This leads to inflammation of these cells and release of chemokines, such as Il-8 (or mouse equivalent MIP-2). Upon pneumococcal infiltration, alveolar macrophages also release IL-8/MIP-2 and TNF-α. The release of IL-8/MIP-2 and TNF-α attracts neutrophils to the infected area. Neutrophils phagocytose pneumococci through complement C3 deposition and its associated receptors (opsonophagocytosis). Cytokines, such as TNF-α, are also released by infiltrating neutrophils, which in turn leads to a further increase in neutrophil infiltration into the infected areas. In the later stages of lung infection the pneumococci are lysed by activation of Lyt-A or by neutrophils, both events cause a release of pneumolysin into the surrounding tissue. This will have a wide range of cytotoxic and inhibitory effects on host tissue and immune cells, including complement activation, inhibition of neutrophil respiratory burst and release of the antibacterial, vasodilatory, nitric oxide from macrophages. In addition, the release of pneumolysin is thought to increases the probability of pneumococcal survival and growth in vivo.

Both the cytotoxic and complement activity caused by pneumolysin plays a role in inducing inflammation. Using pneumococci expressing cytotolytic and/or complement-activating negative pneumolysin to infect mice intranasally, we showed [47] that each of these bacteria resulted in less severe pneumonia and bacteraemia compared to wild-type infected mice. A recent study has shown that cytolytic and complement activating activities of pneumolysin were responsible for accumulation of different patterns of leucocyte influx [25]. It was hypothesized that the complement activation by pneumolysin was mainly responsible for T cell influx into lung tissue, whereas the cytolytic activity was responsible for neutrophil recruitment [25].

The mechanisms of T cell involvement in immune responses to pneumococci were generally poorly understood. We have shown, however, that T lymphocytes were involved at an early stage in the host immune response to pneumococcal infection, in the absence of antipneumococcal antibody, whereby a significant early accumulation of T cells whose peak of infiltration in the lungs during intranasal pneumococcal infection of mice in vivo was coincident with the phase when pneumococcal growth ceased [44]. The maximum accumulation of T cells was found to be significantly less intense and delayed in the absence of the pneumococcal toxin pneumolysin [44]. Furthermore, in vivo mouse infection studies verified that both the cytolytic and complement-activating activities of pneumolysin contributed to the influx of inflammatory cells such as neutrophils. However, only the complement-activating role of the toxin was involved in the recruitment of T cells to inflammatory lesions and not its pore-forming activity [25]. A very recent study has just shown that MHC-II knock-out mice, which were shown to be CD4 T cell-negative, were significantly more susceptible to pneumococcal bronchopneumonia and septicaemia than their isogenic CD4 sufficient wild-type parents [48]. Indeed, MHC-II-deficient mice developed pneumonia and bacteriaemia at a significantly earlier stage than wild-type mice and had significantly shorter survival times [48].

Another interesting recent development has been the discovery of apparent involvement of Toll-like receptors (TLRs) in pneumococcal disease. TLRs have a crucial role in recognizing structurally conserved pathogen-associated molecular patterns, hence allowing immediate responses to limit or eliminate invading microorganisms. It has been shown that cell wall components of the pneumococcus are recognized by TLR-2 [49], and recent preliminary evidence would also suggest that the inflammatory response of murine macrophages to the pneumococcal toxin pneumolysin depends on TLR-4 and that mutant mice that lack TLR-4 are significantly more susceptible to invasive disease and death following infection with wild-type pneumococci compared to control mice [50]. The suggestion that TLR-4 and TLR-2 are involved in the host innate immune response to pneumococcal infection is of great interest, and hints at the important protective role these receptors may have during host response to infection.

THE ROLE OF PNEUMOLYSIN IN PNEUMOCOCCAL MENINGITIS

Pneumococcal meningitis causes many alterations in neuronal function. Clinically this is manifested as coma in the acute disease, and in the longer term sequalae include mental retardation, learning disabilities and focal neurological deficits [51]. Gross meningeal inflammation has been associated with blood vessel inflammation, central nervous system (CNS) necrosis, neuronal loss and general inflammation of brain tissue and cranial nerves [52]. Animal models of pneumococcal meningitis have been pivotal in our understanding of the mechanisms of the disease. Indeed, improvement of current antibiotic and adjunctive therapeutic strategies to treat pneumococcal meningitis rely on good animal models. Using animal studies other neurological disturbances have been identified, including alterations of cerebral blood flow [53] intracranial hypertension [54], cerebrospinal fluid (CSF) hydrodynamic alterations [55] and the development of brain oedema [56]. It is generally believed that prior to meningitis the pneumococci must first successfully colonize the circulation, and providing there is no trauma, breach the blood–brain barrier before they are able to colonize the CSF. This sequence of pathogenesis has been challenged recently by evidence of transport of pneumococci directly from the nasopharynx into the CNS via the olfactory neurones [57]. Once in the CSF, three stages of events in pneumococcal meningitis have been identified. First, invading bacteria cause an inflammatory response in the CSF; the cerebrovasculature then reacts to this inflammation resulting in hyperaemia and ischaemia, and finally toxic events in the brain cells lead to permanent brain injury [58]. The role of pneumolysin in bacterial meningitis in these events is discussed below. Other recent reviews on the pathogenesis of pneumococcal meningitis have focused on the mechanisms of neuronal injury [59,60].

There are only a limited number of reports in the literature of the effect of pneumolysin in in-vivo models of pneumococcal meningitis. The first study on the effects of pneumolysin in vivo reported no difference in inflammation in the brains of rabbits infected with pneumolysin-deficient or wild-type (pneumolysin-containing) pneumococci [61].

However, several recent reports have not been in agreement with this conclusion. A recent study has shown that when mice were infected directly into the brain with various virulence mutants of pneumococci, including neuraminidase- and hyaluronidase-deficient mutants, only the pneumolysin-negative mutant (PLN-A) showed attenuated meningitis [62]. The study also showed that the mice infected with pneumolysin-deficient pneumococci had a reduced sepsis. All the infections in this study resulted in at least a small degree of hippocampal neuronal damage, but it was more severe with wild-type bacteria [63]. The hippocampal neurones also underwent apoptosis [5] in rabbits infected with wild-type pneumococci in a pneumolysin dependent fashion. These data [5,62] are consistent with the findings from our laboratory that meningitis symptoms and ependymal damage were attenuated in rats cisternally infected with pneumolysin-deficient pneumococci (unpublished). Therefore, conflicting evidence for a role of pneumolysin in pneumococcal meningitis exists. The initial study, suggesting a limited role for pneumolysin in pneumococcal meningitis [61], contrasts with a subsequent body of evidence which has shown that pneumolysin is an important virulence determinant in meningitis [5,62]. It is imperative that further studies are performed in order to determine the precise role of pneumolysin in pneumococcal meningitis.

The passage of pneumococci from the blood into the brain is generally the accepted route by which meningitis develops. Using a microvascular endothelial cell culture model it was shown that pneumococci expressing pneumolysin were able to breach the endothelial cells, whereas mutant pneumococci deficient in pneumolysin were unable to penetrate the cell barrier [64]. This is good evidence that pneumococci expressing pneumolysin may penetrate readily the vasculature of the central nervous system and gain direct entry into the CSF. Because the vascular epithelial cells regulate the passage of molecules from blood to CSF and the ciliated ependyma has a similar function in the regulation of molecules from CSF to brain, it is reasonable to hypothesize that the effect of pneumolysin on each cell type will share common characteristics. We have shown that pneumolysin is toxic to ciliated ependymal cells [12,13]. In addition, we have shown that pneumolysin released from pneumococci inhibited ependymal ciliary beat frequency [11] and that co-incubation with pneumolysin antibodies that block toxin binding and pore formation also blocked the inhibition of ependymal ciliary beat frequency (unpublished). The findings from these ex vivo studies are entirely consistent with the damage to the ependyma observed following cisternal infection of the CSF of the rat with wild-type pneumococci [65].

CONCLUSION

Pneumococcal diseases are clearly mediated by host and bacterial factors, which lead ultimately to cellular and tissue damage and death of the host. The evidence from animal infection studies points clearly to an integral role for pneumolysin in invasive pneumococcal diseases. Blockade or neutralization of pneumolysin thus appears to be a potentially important approach for new therapeutic interventions to treat pneumococcal diseases. It is essential that we characterize fully the role of pneumolysin in the molecular pathogenesis of pneumococcal infections so that suitable vaccines and therapies can be developed.