Main

For 100 y, clinicians have asked how Streptococcus pneumoniae kills the human host. This inquiry has revealed many elements of our understanding of the basic biology of infection (Table 1)[reviewed in Watson et al.(1)]. Bacterial capsules enhance virulence and anticapsular antibodies are protective against disease, providing the first polysaccharide-based vaccine. Leukopenia enables noncapsulated pneumococci to cause progressive disease. Formation of a capsule is an inheritable trait and the “transforming principle” was identified as DNA. Efforts to kill pneumococci have suggested the mechanism of action of penicillin and raised the spectre of antibiotic tolerance and resistance. Only in the last 15 y, however, has a well rounded hypothesis been formulated for how this pathogen kills the human host. The Gram-positive cell wall has been dissected in structure and bioactivity and found to be a major source of inflammatory components. This insight suggested a rationale for improving the outcome of disease by controlling exaggerated host responses to the cell wall, a theory that is gaining practice in the setting of the most devastating infection, meningitis. Finally, with the application of molecular biology and genomics, the most recent insights have led to an understanding of the interaction between the pneumococcal surface and the human cell surface: the molecular mechanism of infection [reviewed in Tuomanen et al.(2) and Cundell et al.(3)]. This suggests that the century of struggle to attenuate the impact of this pathogen may finally close with the formulation of a pneumococcal protein-based vaccine with efficacy for all ages. The three major pathogens of children, the pneumococcus, the meningococcus, and Haemophilus influenzae, share a provocative similarity of physiology: they cause meningitis, they are naturally transformable, and they lyse and die in stationary phase. Understanding the pneumococcus may reveal why this master plan is so suited to invasive disease in children.

Table 1 Contributions of pneumococcal research to biomedical science
Full size table

ADHERENCE, INVASION AND VIRULENCE

Pneumococci bind avidly to cells of the upper and lower respiratory tract (Fig. 1). Attachment does not involve fimbriae or invoke actin polymerization as has been seen for other invasive bacteria. Rather, adherence is a quiet process enabling carriage for several weeks without an overt inflammatory response, and yet invoking serotype-specific immunity based on the structure of the capsular polysaccharide of the infecting strain. In the case of symptomatic disease, however, adherence proceeds to bacterial internalization into host cell vacuoles by receptor-mediated endocytosis(3). How physiologic endocytosis is corrupted by the pneumococcus such that the bacteria-vacuole complex transits the cytoplasm and extrudes the bacteria on the opposite side of mammalian cells is unknown. However, after a complex series of events taking up to 10 h, pneumococci pass from the respiratory mucosa to the blood stream and beyond.

Figure 1
figure 1

Pneumococci adherent to an A549 lung epithelial cell. Scanning electron micrograph made by K. McDonough and M. Florczyk, New York State Department of Health, Albany, NY.

Full size image

Like most bacteria, adherence of pneumococci to human cells is achieved by presentation of surface proteins that bind to at least five different eukaryotic carbohydrates in a lectin-like fashion(4). Pneumococci bind to at least two different ligands on noninflammed pulmonary epithelium and vascular endothelium. These two resting cell surface sugar specificities can be viewed as the molecular targets involved in the asymptomatic presence of pneumococci in the lung and vascular space. In experimental models, low levels of pneumococci have been shown to persist on the mucosa of the alveolar space without overt disease(5). The incidence of bacteremia resulting from pneumococcal pneumonia is less than 1 in 100. Similarly, the incidence of invasion of the meningeal space, even in the face of high grade bacteremia, is rare. It has been proposed that the conversion to invasive disease involves the local generation of inflammatory factors, for instance by intercurrent viral infection, which change the number and type of receptors available by activating human cells(6). Presented with an opportunity in this new setting, pneumococci appear to take advantage and engage one of these up-regulated receptors, the PAF receptor(6). Within minutes of the appearance of the PAF receptor, pneumococci undergo waves of enhanced adherence and invasion. Pneumococci fail to invade resting cells(0.1%). However, 2-3% of the inoculum moves to an intracellular vacuole in activated cells as demonstrated by gentamicin protection studies(6) as well as electron microscopy(7). The binding to activated cells is inhibitable by PAF receptor antagonists and by sugars, such as the human milk component lacto-N-neotetraose.

How the PAF receptor-dependent uptake of pneumococci leads to transcellular migration and exit of living bacteria at the albumenal surface remains to be studied in detail. Results in the rabbit pneumonia model suggest this transmigration, at least in part PAF receptor-dependent, is a critical step in advancing disease(6). Thus, the participation of the host in an inflammatory response, in this case up-regulating the presentation of the PAF receptor, is an important element contributing to development of symptomatic, invasive infection. It also suggests that pneumococcal elements required to negotiate this step will be key virulence determinants operative more frequently in virulent strains and most likely subject to regulated expression.

A Model for Pneumococcal Adherence

The nature and multiplicity of the adhesive determinants of pneumococci reflect the complexity of attachment. Genetic and biochemical evidence indicates that each human cell sugar target has at least one dedicated pneumococcal ligand, i.e. there are at least five specific adhesins. Analysis of mutants that fail to adhere indicates adhesins can be globally or individually inactivated by pneumococcal regulatory elements. Some mutants defective in autolysis or DNA transformation are also deficient in adherence, suggesting that there may be cross-regulation of important events in pneumococcal physiology.

Appreciation that important events in pneumococcal physiology were interrelated provided the key to the development of a model of pneumococcal adherence and virulence consistent with the surprisingly large number of potential participating elements(8). Temporal analysis of the wild type pneumococcal life cycle is shown in Figure 2. Major events appear to be regulated with time in accordance with a quorum sensing paradigm. Transformation occurs in a brief window during early log phase coincident with the extracellular appearance and disappearance of the 17-amino acid activator substance (competence stimulating peptide)(9). Transformation is followed by a transient ability to adhere to the PAF receptor and subsequently to sialic acid. These transient binding capacities are distinct from the constitutive ability to bind to resting cell receptors of the lung and vascular endothelium. Finally, in the stationary phase, autolysis is triggered. Supernatant fluids collected at the peak of each activity can confer that activity on cells from other points in the life cycle consistent with cell signaling triggered by an autoinducer(s) produced at discreet cell densities. This model suggests that, for transformation, autolysis, and two types of adhesion, there exist extracellular autoinducers that indicate four different cell densities. For each of these four checkpoints, a signaling cascade exists that triggers expression of a regulon, resulting in the coordinate appearance of many gene products required to enact each physiologic event.

Figure 2
figure 2

The pneumococcal life cycle. Pneumococci pass through four distinct phases during logarithmic growth. Each is triggered at a specific cell density consistent with a quorum sensing event. Natural DNA transformation: At low bacterial cell density, production of the 17-mer activator peptide is triggered and induces expression of over a dozen genes required from the binding, uptake, and incorporation of exogenous DNA .Adherence to PAF receptor of activated cells: Although adherence to resting human cells is expressed throughout the growth cycle at the same intensity, adherence to the PAF receptor and subsequently to sialic acid occur in transient, overlapping peaks. The extracellular signals for these events are unknown. Finally in stationary phase, pneumococci autolyze through the activity of the endogenous murein hydrolase, the N- acetylmuramyl-L-alanine amidase. The enzyme is triggered by an unknown extracellular product of <3 kD size. For details, see Cauwels et al.(16).

Full size image

Regulation of Adherence

The opacity locus. Pneumococci spontaneously vary colonial morphology at high frequency between opaque and transparent colony variants(10). Opaque colonies fail to cause disease in a rat intranasal challenge model, whereas transparent variants colonize the nasopharynx and cause bacteremia. A genetic locus conferring opacity has been identified, although its function has not been clarified(11). Although opaque and transparent pneumococcal variants adhere to a similar degree to nonactivated epithelial and endothelial cells, enhanced adherence to cytokine-stimulated cells or PAF receptor transfected COS cells is limited to transparent variants(12). This is consistent with the advantage shown by the transparent pneumococci in producing invasive disease in vivo. Transparent pneumococci are also transformation-deficient and exhibit reduced autolytic properties. Recent studies indicate that the pattern of surface proteins on the two phenotypes differs, suggesting a class of proteins is regulated together with phase variation.

Bacterial permeases. Peptide permeases capture and transport small peptides or amino acids from the environment into the cell. They are excellent candidates for the transport of quorum sensing molecules, and genetic experiments suggest this is the case for pneumococci(13). Among this class of proteins is the recently described surface protein PsaA(14). Loss of function of PsaA affects a wide range of adhesive events, but PsaA is not itself a structural adhesin(15).

Two-component signal transduction systems. Most bacteria sense their environment through two-component signal transduction systems. Binding of a ligand to a surface-exposed histidine kinase triggers phosphorylation of a cognate response regulator, which in turn moves to the DNA and changes gene transcription. Three two-component systems have been characterized in pneumococcus, and they appear to cross-link regulation of transformation, autolysis, and virulence(16–18). Analysis of“who speaks to whom about what” is still underway, but the loss of function of any two-component system has pleotropic affects crippling the ability of pneumococci to proceed through the timely development of transformation, adherence, and autolysis. These studies justify suggesting that pneumococci undergo a program of development with all members of a culture community behaving sequentially in a proscribed, similar manner.

Choline Binding Proteins: Candidate Structural Genes

Several laboratories have described a family of surface proteins bound to the choline component of the cell wall teichoic acid or lipoteichoic acid of pneumococci). A signature choline binding domain was discovered and fully characterized by Lopez in his studies of the autolytic enzyme(19). Other proteins containing this domain include the autolysin of the pneumococcal phage and the protective antigen, pneumocococcal surface protein A (PspA)(19, 20).

Twelve additional choline-binding proteins have been identified in our laboratory(21). These proteins are surface-exposed and react with convalescent human antisera. When copurified, these proteins can inhibit pneumococcal adherence to endothelial and epithelial cells, suggesting that several may contribute to adherence. Antiserum to the mixture of proteins is protective against pneumococcal challenge in rats. A mutant defective in one choline-binding protein, CbpA, fails to colonize the nasopharynx. Thus, the choline-binding protein family of surface elements contains the structural gene for autolysis (amidase), at least one candidate adhesin (CbpA), and a protective antigen (PspA). Recent evidence suggests that a choline binding protein may also stabilize the activity of the activator substance for transformation(22). Variation in the expression of choline-binding proteins during opaque and transparent phase switching is consistent with the multiple changes that occur in bacterial physiology with this transition(10). Choline-binding proteins may serve as a family of structural genes carrying out much of the important cellular physiology of the pneumococcus (Fig. 3).

Figure 3
figure 3

Choline binding proteins as the surface effectors of pneumococcal physiology. The cell wall (shown as a network) bears teichoic acid (long ovals) substituents which are decorated by multiple choline adducts (P-cho). P-cho supports the reversible binding of a set of at least 12 choline binding proteins (shown as caps). These proteins influence transformation, adherence, and autolysis, the major events in the pneumococcal life cycle. Changes in these proteins occurs spontaneously during phase variation from transparent to opaque.

Full size image

The Cell Wall Choline as an Adhesive Ligand

An unusual feature of the pneumococcal cell wall structure is the presence of phosphorylcholine in the teichoic acid and lipoteichoic acid(23). Pneumococcal choline-bearing cell wall structures appear to serve several functions (Fig. 3). First, they are a platform for docking of the choline-binding protein family on the bacterial surface. Second, choline potentially interacts directly with the PAF receptor during bacterial adherence. Finally, choline-mediated binding of some soluble cell wall fragments to human cells induces signaling in the host cell, resulting in generation of some elements of the acute phase response (see below). Thus a significant amount of pneumococcal pathophysiology is focused on the cell wall choline (at least as a determinant of bacterial binding and migration across cells and as a key inflammatory mediator). That such choline-related biology might extend to other respiratory pathogens is an exciting possibility that has been suggested by two findings: choline decorates the surface of the polar lipid of Mycoplasma pneumoniae(24), and the presence of a choline adduct on endotoxin correlates with phase variable, serum sensitivity of H. influenzae(25). Definition of the genes involved in placement of choline on the surface of these pathogens will likely reveal important convergent evolution of pathogens capable of targeting the choline-rich environment of the human lung.

MECHANISMS OF PNEUMOCOCCAL INFLAMMATION

The Cell Wall as a Library of Inflammatory Fragments

When cell wall, cytoplasm, and capsule are compared for inflammatory capacity, cell wall has the highest specific activity(26, 27) (Table 2). This activity is not shielded by overlying capsule on the native bacteria. The signs and symptoms of infection induced by cell wall mimic those of living bacteria in animal models of meningitis, pneumonia, and otitis media(27–29). This bioactivity is influenced by both the structure of individual cell wall components and their release in soluble form from the bacteria into the host milieu. For instance, the major building block of the peptidoglycan portion of the pneumococcal cell wall is a potent stimulus of blood-brain barrier permeability when injected alone i.v. in rabbits(30). The release of this component varies according to the amount of cell wall remodeling that proceeds along with bacterial growth. Clinical strains and their isogenic laboratory derivatives that have defects in release of cell wall fragments induce an attenuated pattern of disease(31). Thus, cell wall structure and release are major determinants of the course of disease.

Table 2 Bioactivities of pneumococcal cell wall components
Full size table

Teichoic and lipoteichoic acids contribute strongly to host defense responses associated with acute inflammation. They activate the alternative pathway of the complement cascade, bind the acute phase reactant C-reactive protein, activate procoagulant activity on the surface of endothelial cells, induce cytokines and PAF upon binding to epithelia, endothelia, and macrophages, and initiate the influx of leukocytes(7, 32–38). The IL-1 response generated by the pneumococcal cell wall is particularly strong, exceeding that for endotoxin at least 10-fold on a bacterial colony forming unit basis. In contrast, induction of TNF requires >100 times more cell wall than the induction of IL-1, even in the presence of putative serum binding components(35, 36). Some of these effects arise through the interaction of cell walls with CD14, a cell surface receptor known to initiate the inflammatory cascade for endotoxin(36, 39). Other receptors are suspected also to participate in cell wall-induced events, because mice resistant to endotoxin/CD14 effects by virtue of knockout mutations in the TNF receptor or p50 of NF-κB still die of pneumococcal infection(40, 41). Conversely, mice deficient in ICAM-1 have a poorer prognosis for Gram-negative than for Gram-positive meningitis(42).

Translation into Medicine: Attenuation of Disease

Intervention at the level of adherence. It has long been suggested that understanding bacterial adherence would generate excellent vaccine candidates and novel therapeutics. Clearly many of the proteins described above will be analyzed as potential vaccine candidates with the aim of finding a pneumococcal protein which could extend the protective effects of vaccines to most serotypes and most age groups. To examine possible therapeutics based on anti-adherence strategies (Fig. 4), both the PAF receptor antagonist and a series of anti-adhesive oligosaccharides were tested for the ability to attentuate the course of pneumococcal disease, i.e. pneumonia in a rabbit model and colonization of the nasopharynx in a rat model(6, 43). The PAF receptor antagonist or lacto-N-neotetraose eliminated pneumococci from the nasopharynx and lungs over a period of 48 h and resulted in protection from bacteremia. These agents also ameliorated established pneumonia. The PAF receptor antagonist and the anti-adhesive oligosaccharides are promising agents for interrupting carriage and transmission of pneumococci, particularly in view of the current interest in limiting the use of antibiotics in the absence of overt disease so as to decrease the emergence of resistance.

Figure 4
figure 4

Anti-adhesive interventions for pneumococcal disease. Adherence to resting cells is representative of asymptomatic colonization that can be blocked by excess soluble sugar (1) of several different structures (sticks). Adherence to activated cells involves two specificities, lacto-N-neotetraose (LNnT) and PAF receptor (PAF r), and is followed by invasion to cause bacteremia. This can be blocked by excess LNnT (3) or PAF receptor antagonist (2). Adapted from Cundell et al.(6) and Idanpaan-Heikkila et al.(43).

Full size image

Intervention at the level of cell wall induced inflammation. Given the complex relationships between the participants in the acute inflammatory response to pneumococci, it is not surprising that a single critical agent has not been identified that can “shut off” the disease. Attenuating the acute host response, however, has direct clinical application to improving the outcome of disease. A major change in the therapy of meningitis arose from the observation that pneumococcal cell wall pieces are bioactive when released from growing bacteria and, even more so, from bacteria undergoing antibiotic-induced lysis. This provided an opportunity to mitigate cell wall-associated damage by down-modulating the host response during antibiotic therapy(44, 45). Over the first few hours of antibiotic therapy, the leukocyte density in cerebrospinal fluid can increase one to two orders of magnitude(46). This burst is sufficient to injure host tissues as evidenced by the significant attenuation of damage upon the inhibition of leukocyte recruitment(44, 45). The use of steroids during the early phase of antibiotic therapy to inhibit this response has recently been deployed in the clinical setting of childhood meningitis(46). As a greater understanding of the interplay between mediators of inflammation from the pneumococcus and the host is obtained, adjuvant therapies that more cleverly tailor the down-modulation of antibiotic-induced inflammation will be found such that killing the bacteria will more readily also save the patient.