Pneumococcal Vaccines

ABSTRACT

Streptococcus pneumoniae is a Gram-Positive pathogen that is a major causative agent of pneumonia, otitis media, sepsis and meningitis across the world. The World Health Organization estimates that globally over 500,000 children are killed each year by this pathogen. Vaccines offer the best protection against S. pneumoniae infections. The current polysaccharide conjugate vaccines have been very effective in reducing rates of invasive pneumococcal disease caused by vaccine type strains. However, the effectiveness of these vaccines have been somewhat diminished by the increasing numbers of cases of invasive disease caused by non-vaccine type strains, a phenomenon known as serotype replacement. Since, there are currently at least 98 known serotypes of S. pneumoniae, it may become cumbersome and expensive to add many additional serotypes to the current 13-valent vaccine, to circumvent the effect of serotype replacement. Hence, alternative serotype independent strategies, such as vaccination with highly cross-reactive pneumococcal protein antigens, should continue to be investigated to address this problem. This chapter provides a comprehensive discussion of pneumococcal vaccines past and present, protein antigens that are currently under investigation as vaccine candidates, and other alternatives, such as the pneumococcal whole cell vaccine, that may be successful in reducing current rates of disease caused by S. pneumoniae.

With the discovery of the pneumococcus in 1881, it became apparent that this Gram-positive pathogen was a major cause of serious and often fatal pneumonia (1). It is also a major cause of meningitis and otitis media in children. Pneumococci are the largest cause of community-acquired pneumonia in the developed world. According to the Centers for Disease Control and Prevention, in the United States the combined rates of invasive pneumococcal diseases has shown a decline since the introduction of the conjugate vaccines, with rates among the elderly declining from 59/100,000 in 1998 to 23/100,000 in 2015 and rates among children under 5 years of age falling from 95/100,000 in 1998 to 9/100,000 in 2016 (https://www.cdc.gov/pneumococcal/surveillance.html). The rate of meningitis in children in the United States is about 4 cases per 100,000 children, with a fatality rate of about 15% (2). In the developing world pneumococci are an important cause of childhood deaths due to bacterial respiratory infection following viral disease. The World Health Organization suggests that globally, such infections have been reported to have killed an estimated over 500,000 children in 2008 (http://www.who.int/immunization/monitoring_surveillance/burden/estimates/Pneumo_hib/en/). Recently, about one-third to one-half of pneumococci recovered from humans in the United States have been found to be at least partially resistant to penicillin, and penicillin-resistant strains are frequently also resistant to other common antibiotics (3). The rise of antibiotic resistance among pneumococci has already complicated treatment, especially of meningitis (4), and threatens to greatly increase the morbidity and mortality caused by pneumococci unless new control measures are developed.

It has long been recognized that the best management of most infectious disease is prevention. Vaccines offer the prospect of a highly cost-effective means of preventing morbidity and mortality caused by pneumococci. This article provides a concise summary of issues critical to the development and application of pneumococcal vaccines. There are several relatively recent reviews that address this topic in more detail (5–11).

In the preantibiotic era, vaccination attempts utilized whole killed pneumococci injected parenterally. Although such vaccines were sometimes protective in humans, they were also highly reactogenic. These killed vaccines were mainly used to elicit antibody in animals for passive treatment of infected humans (12). In 1933 it was clearly demonstrated that antibody to type-specific capsular polysaccharides (PSs) could be highly protective (13). However it soon became apparent that different strains of Streptococcus pneumoniae each expressed one of many different PSs. Most subsequent vaccine attempts focused on the use of mixtures of the isolated PSs to elicit protection (7, 14). However, the inability of young children to make adequate responses to most soluble PSs led to the eventual development and licensing of an immunogenic PS-protein conjugate vaccine for children (6, 15). However, the wide diversity of pneumococcal PS types, the failure of PS vaccines to protect against strains with PSs not in the vaccines, and the complexity and expense of highly valent conjugate vaccines have led to investigations of other much more cross-reactive pneumococcal antigens (primarily proteins) as potential vaccine candidates.

PNEUMOCOCCAL CAPSULAR PS VACCINE

During the 1930s, capsular PS was shown to be able to elicit protective antibodies. At that time it was generally assumed that capsular PS was the target of all, or most, protective antibodies (1, 13). Currently, over 90 distinct serotypes of capsular PS have been identified for pneumococci (16). Each strain can make only one PS, and for a strain to make a different PS requires a genetic transformation with DNA of another strain. Only about one-third of the greater than 90 serotypes occur with a significant frequency in adult infections, and the distribution of common serotypes is even more restricted in children (6, 17–20). Following the identification of capsular PSs as protection-eliciting molecules, vaccines comprising relatively pure PSs were used successfully in nursing homes and among military personnel in the 1940s. However, during this time antibiotics were becoming widely available, and these drugs diminished the perceived impact of pneumococcal infections. The two commercially available pneumococcal vaccines were withdrawn from production for lack of demand (7).

Despite a growing selection of antibiotics and increasingly sophisticated critical care technology, morbidity and mortality from invasive pneumococcal infections have remained high. Antibiotics and supportive care alone are not sufficient to completely eliminate the impact of pneumococcal disease because in many cases deaths occur so quickly that antibiotics are not effective (21). During the 1970s, renewed interest in pneumococcal vaccines (14) led to clinical trials of a PS vaccine in adults at high risk for invasive pneumococcal disease (IPD) in South Africa and New Guinea. Based largely on encouraging results from these trials, a 14-valent PS vaccine was licensed in the United States in 1977. The vaccine was expanded to 23 PS serotypes in 1983 and is licensed for use in adults in the United States (6, 7, 22).

Clinical trials examining the efficacy of the PS vaccine continued after licensure, with somewhat mixed, and controversial, results (7). A significant impediment to conducting the most rigorous controlled clinical trials has been the ethical concern of denying at-risk groups an approved and recommended vaccine. A recent systematic review and meta-analysis to examine the efficacy of the 23-valent vaccine (PPSV23) in adults >60 years of age, has shown a pooled vaccine efficacy of 73% (95% confidence interval [CI]: 10 to 92%) against IPD in clinical trials, 45% (95% CI: 15 to 65%) in cohort studies and 59% (95% CI: 34 to 74%) in case control studies (22). The higher efficacy of former trial data, with a 2.5-year follow-up, compared to the lower efficacy seen in the observational cohort and case control studies, with a 5-year follow-up, may reflect a waning protection over time (22, 23). Although the PS vaccine is cost-effective even when considering only bacteremic disease, this vaccine is not generally accepted to reduce the incidence of pneumococcal pneumonia or localized upper respiratory infections such as acute bronchitis, sinusitis, or otitis media (24).

A major drawback to the PS vaccine is the poor immunogenicity of PS antigens in children and elderly adults, primarily due to the vaccine eliciting a T cell-independent immune response (25). The response to PS varies by age and with the presence of certain underlying chronic medical conditions, most notably, infection with human immunodeficiency virus (HIV). Certain capsular PS serotypes in the 23-valent vaccine, such as types 6B, 9V, 19F, and 23F, induce relatively low amounts of specific antibody, which fall to prevaccination levels within 3 years (26). In addition to a reduced quantity of antibody, a significant number of elderly adults have impaired functional antibody responses to PS as defined by in vitro avidity and opsonophagocytosis assays (27, 28).

POLYSACCHARIDE-PROTEIN CONJUGATE VACCINES

Children less than 2 years of age are at increased risk for infections caused by encapsulated bacteria; consequently, this group had the highest incidence of invasive pneumococcal infections in the preconjugate era (27). Children in this age group do not consistently respond to PS antigens, which directly cross-link surface immunoglobulin on B cells and do not have an absolute requirement for T cell help (29–31). Whatever the underlying mechanisms, children less than 2 years old respond poorly to many of the common pneumococcal capsular types (32), and alternatives to purified PS antigens are needed to protect young children.

The ability to make antibodies to protein antigens appears quite rapidly after birth, and T cell-dependent antibody production is highly efficient. The development of PS-protein conjugate vaccines for children takes advantage of their ability to elicit T cell-dependent responses to PS covalently attached to immunogenic proteins. This strategy has proven highly successful in preventing both invasive disease and colonization with Haemophilus influenzae type b (33). The process of synthesizing conjugate vaccines requires chemical reactions, which covalently attach PS to protein; nonconjugated mixtures of peptides and PS do not induce T cell-dependent PS antibodies (34).

Immunization with PS conjugate vaccines induces effective memory antibody responses in infants and children, and antibody levels can be boosted with repeated immunization (35). A large clinical trial in infants using a four-dose schedule of immunization with a heptavalent conjugate vaccine proved this strategy to be highly effective in preventing invasive (bacteremic) disease (30, 36, 37). However, this conjugate vaccine has shown somewhat more modest efficacy in preventing pneumonia (15) and otitis media (38). The heptavalent pneumococcal conjugate vaccine (PCV7) was licensed in 2000 in the United States for use in all children less than 2 years of age and for high-risk children less than 5 years of age (37). Since the vaccine serotypes that colonize and infect children include most of the serotypes associated with antibiotic resistance, and since children carry pneumococci at a much higher frequency than adults, there are expectations that childhood immunization may reduce the transmission of antibiotic-resistant pneumococci in the entire population (39–42). It has been observed, however, that significant replacement carriage and otitis media occur with capsular types not present in the vaccine (38, 39, 42). Since the introduction of the 7-valent vaccine, a significant increase in invasive disease was seen among nonvaccine strains, which led to the introduction of the 13-valent conjugate vaccine in 2010 (43–45). However, despite the introduction of the new conjugate vaccine, strain replacement still persisted in carriage and invasive disease, with rates of invasive disease due to nonvaccine strains ranging from 57.8% in North America to 71.9% in Europe (46–50). Also of concern are observations that immunization with the conjugate vaccine alters the bacterial flora of the upper respiratory tract in children and has affected the pathogens recovered from children with carriage and acute otitis media (51, 52). A potentially worrisome trend has been noted in increased rates of carriage of Staphylococcus aureus in children immunized with the pneumococcal conjugate vaccine (51, 53).

According to 2015 data from the Centers for Disease Control and Prevention, adults over the age of 65 years had the highest risk of IPD in the United States (https://www.cdc.gov/abcs/reports-findings/survreports/spneu15.html). The distribution of pneumococcal serotypes commonly causing invasive disease in adults is much larger than that for children. The Advisory Committee on Immunization Practices recommended the current 13-valent conjugate vaccine for use, in conjunction with the 23-valent PS vaccine, in adults over the age of 65 years (54). This recommendation was based on the results of a large efficacy trial in the Netherlands, which showed 75% and 46% efficacy against vaccine type invasive disease and pneumococcal pneumonia, respectively, among people over 65 years (55). It is recommended that the 13-valent vaccine be administered prior to the 23-valent vaccine, because there is evidence that prior PS vaccine administration reduces the immune response to the conjugate vaccine (56, 57). The CDC is currently evaluating the efficacy of the stand-alone 13-valent conjugate vaccine or the conjugate vaccine in conjunction with the 23-valent PS vaccine in older adults in the United States. The 13-valent conjugate vaccine is also recommended for routine use among people over the age of 19 with immunocompromising conditions, functional or anatomic asplenia, cochlear implants, or cerebrospinal fluid leak (54, 58). Attempts are also underway to develop conjugate vaccines with much higher valence to broaden coverage of different serotypes (59, 60).

Strategies to enhance the immunogenicity of PS have been tested in small animal and human studies, but none have yet reached clinical application. A conjugate of type 6B PS and tetanus toxoid reduces carriage in mice when administered intranasally with the cholera toxin B subunit (61). T cell help during response to protein antigens involves binding of CD40 on B cells by CD154 expressed on T cells. Administration of anti-CD40 antibodies along with capsular PS results in predominantly high-affinity IgG antibodies, which is a hallmark of responses to protein antigens (62). Small amounts of interleukin 12 have been shown to be a potent adjuvant for pneumococcal PS in mice (63). Coadministration of immune modulators may prove to be an effective and practical strategy to enhance immune responses to PS. Unless substantial progress is made with protective protein antigens or protein-PS conjugates, novel methods for adjuvant responses to purified capsular PS will remain an important area of investigation into pneumococcal vaccines. One such approach has been the development of a multiple antigen-presenting system (MAPS), which combines various antigens, including PSs and proteins, in a synthetic acellular system (64). Such complexes of antigens in some ways mimic the pneumococcal whole-cell vaccine (WCV) with probably much less risk of undesired reactogenicity. The MAPS complex is highly efficient in generating a broad immune response, including B and T cell (Th1 and Th17) responses, and is particularly effective in the case of S. pneumoniae, because it circumvents the issue of covering multiple serotypes to avoid the serious problem of serotype replacement. An approach with multiple antigens that collectively protects against carriage, pneumonia, and sepsis, delivered through a mucosal route, would be expected to be optimally protective.

NONCAPSULAR PS VACCINES

Overview

Polyvalent pneumococcal vaccines based on purified PS have been available for over 3 decades, but their clinical efficacy has been limited by poor immunogenicity in high-risk groups (particularly young children). As discussed above, the problem of poor vaccine immunogenicity in children has been addressed by conjugation of the PS to protein carriers, thereby converting the PS from T cell-independent to T cell-dependent antigens. However, serotype coverage remains limited, and it is unlikely that significantly more than 13 to 15 of the 98 currently known serotypes will be included in future conjugate formulations. Nasopharyngeal colonization with S. pneumoniae is a virtual prerequisite for invasive disease. Data from the original conjugate vaccine trials indicated that, although carriage of vaccine types was reduced, the vacated niche was promptly occupied by nonvaccine serotypes known to cause invasive disease in humans (8). Such “replacement” carriage, as well as replacement disease due to nonvaccine serotypes, has occurred to varying extents in virtually all regions where conjugate vaccines have been widely used since the licensure of the original 7-valent formulation in 2000 and continued after introduction of 10- and 13-valent formulations (65–67). In addition, the cost of the conjugate vaccines remains very high; thus, without philanthropic support, their use in developing countries, where the need for effective pediatric vaccines is the greatest, will remain restricted. In view of this, much attention has focused on the possibility of developing vaccines based on pneumococcal protein antigens common to all serotypes (5, 68, 69). Such proteins, being T cell-dependent antigens, are likely to be highly immunogenic in human infants and able to elicit immunological memory. The pneumococcal proteins also have potential as carriers for the PS in the conjugate vaccines. The vaccine potential of various pneumococcal protein antigens are discussed further below.

PspA and PspC

Pneumococcal surface protein A (PspA) is produced by all pneumococci (70). The protein is important for virulence (71, 72); its presence on pneumococci prevents classical pathway complement deposition via C-reactive protein (CRP) and probably anti-PC antibody (72–74). PspA also appears to reduce phagocytosis of pneumococci even in the absence of C′ and antibody (75). Antibodies to PspA, however, strongly enhance complement deposition (73). PspA may also play a role on mucosal surfaces since it blocks killing by apolactoferrin (76). Antibody to PspA reverses this blockage and enhances killing by apolactoferrin (76). Although serologically variable when examined with monoclonal antibody, PspA is highly cross-reactive when examined with polyclonal sera (70, 77). Monoclonal and polyclonal antibodies to PspA can passively protect mice from otherwise fatal bacteremia and otherwise fatal sepsis caused by pneumococci (78, 79). Parenteral immunization with PspA has been able to protect mice against fatal infections, and this protection has been observed to be highly cross-protective regardless of PspA type (79–81). Intranasal and transdermal immunization with PspA has also been shown to protect against nasopharyngeal carriage in an adult mouse carriage model, raising the possibility that vaccines could be developed to prevent carriage and transmission of pneumococci in addition to invasive disease with pneumococci (45, 61, 82–86). A group in Japan has used PspA for the development of a mucosal vaccination protocol which does not use an immunogenic adjuvant or molecules that can follow the nerves from the nose into the brain. Their protocol and the transdermal route could be particularly important in moving PspA and other mucosal immunogens into humans as the vaccines routes able to elicit both mucosal and systemic immune responses (83, 84).

A number of pneumococcal proteins, including PspA, PspC, and LytA, are able to bind choline and share similar, and sometimes indistinguishable, choline-binding domains near the C-terminal ends of the proteins (87–90). The N-terminal end of PspA is composed of a largely coiled-coil α-helical sequence (89, 91) that is responsible for most of the cross-protective immunity elicited by PspA (92, 93). Paradoxically, this region also contains significant structural and serologic variability within the PspA molecule (92, 94). In spite of its variability at the amino acid sequence level, PspAs are very cross-reactive (70, 79) and can be divided into families based on their cross-reactivity and amino acid sequences (77, 94). Ninety five percent or more of clinical isolates belong to families 1 and 2 (77, 94, 95). Immunization of humans with a single recombinant family 1 PspA leads to antibodies that are able to protect mice from otherwise fatal infection with strains expressing either family 1 or family 2 PspAs (80). However, other data indicate that the best vaccine would probably contain PspAs from both family 1 and family 2 (81, 93, 96). In addition, it has been shown that the 80-120 amino acid long proline-rich domain in the middle of PspA, is also able to elicit protective antibody (97). The use of well-chosen alpha-helical and proline-rich domains in immunizing PspA molecules should ensure coverage of diverse PspAs.

The serum of virtually all adults, and most children over 7 months of age contains detectable antibody to PspA (98, 99). The levels of antibody are higher in adults than in children, and it is thus possible that natural antibodies to PspA will contribute more to the immunity to pneumococcal infection in adults than in young children. It is assumed that immunization to elicit high levels of antibodies to PspA will be able to enhance protection of young children and adults with waning immunity from infections with pneumococci.

PspC is a protein that has some similarity to PspA in its proline-rich and choline-binding domains (100). The gene for this protein was originally identified through its close similarity to PspA (101). However, the alpha-helical domain of PspC is more complex than that of PspA and is present as several very distinct alleles with distinct combinations of functions (102, 103). This protein has been independently discovered by others based on its ability to bind secretory IgA (SpsA) (104), choline (CbpA) (105), and factor H (Hic) (106). It has potential roles in colonization, adherence, and invasion (103–105, 107, 108). Immunity to PspC is able to protect against pneumococcal infection and carriage (103, 108). Recent studies have shown that the NEEK motif of PspC (CbpA) is able to bind to laminin receptors on the blood-brain barrier and thus is able to elicit protection against fatal pneumococcal infection and is suspected to be particularly important in protection against meningitis (109, 110).

Pneumolysin

All pneumococci produce pneumolysin, a potent 53-kDa thiol-activated pore-forming cytolysin. Pneumolysin is released from pneumococci especially during autolysis. This cell-free pneumolysin can attack any cell that has cholesterol in its plasma membrane (111). In addition to its cytotoxic properties, cell-free pneumolysin is capable of directly activating the classical complement pathway away from the pneumococcal surface and does so in the absence of specific antibody. Classical pathway activation by pneumolysin reduces the local concentration of reactive C3 and other classical pathway complement components. As a result, there is a local decrease in serum opsonic activity (112).

Structure-function analysis of pneumolysin has indicated that a domain toward the C terminus of the toxin (amino acids 427 to 437), which includes a unique cysteine residue, is critical for cytotoxicity (113). This cysteine motif is highly conserved among other members of the thiol-activated cytolysin family. Several single-amino-acid substitutions within this region (and other regions involved in cell binding and pore formation) reduce the cytotoxicity of pneumolysin by up to 99.9%. A separate region, which has a degree of amino acid homology with human CRP, is responsible for IgG binding and complement activation, and a mutation in this domain (Asp385→Asn) interferes with both properties (114).

Pneumolysin has a variety of detrimental effects on cells and tissues in vitro, which provide clues to its role in the pathogenesis of pneumococcal disease (111, 115). Complete inactivation of the pneumolysin gene in either a type 2 or type 3 pneumococcus has been shown to reduce virulence for mice challenged by both the intranasal and intraperitoneal routes (116, 117). Compared with the wild-type strain, intranasal challenge with the pneumolysin-negative pneumococci resulted in a less severe inflammatory response, a reduced rate of multiplication within the lung, a reduced capacity to injure the alveolar-capillary barrier, and a delayed onset of bacteremia (118, 119). Pneumolysin also plays a critical role after bacteremia has developed, by modulating the elicited inflammation to reduce its ability to keep bacteremia in check (120). Pneumolysin’s ability to modulate inflammation may be related to its ability to stimulate Toll-like receptor 4 (121) and may contribute to chemotaxis of CD4 T cells (122). By using S. pneumoniae derivatives, in which the wild-type pneumolysin gene was replaced by mutated genes encoding toxins with point mutations affecting either or both of the cytotoxic and complement activation properties, it has been possible to confirm distinct roles for the two toxin activities in the pathogenesis of pneumococcal pneumonia (123–125).

These studies established the importance of pneumolysin in the pathogenesis of pneumococcal disease and identified it as a target for vaccination. It has been known for many years that immunization with purified pneumolysin protects mice against challenge with highly virulent pneumococci (126). Potential problems of toxicity have been overcome by site-directed mutagenesis of regions of the toxin essential for cytotoxicity and complement activation, as described above. Genes encoding these recombinant pneumolysin toxoids (pneumolysoids) have also been inserted into Escherichia coli expression vectors, permitting large-scale production of antigens at low cost (127). Sequence analysis of pneumolysin genes from a wide range of S. pneumoniae serotypes has confirmed that there is very little variation in primary amino acid sequences (>99% identity) (128, 129), so a single vaccine antigen should provide coverage against all pneumococci regardless of serotype. Indeed, immunization of mice with a pneumolysoid carrying a Trp₄₃₃-Phe mutation resulting in a >99% reduction in cytotoxicity (designated PdB) provided a significant degree of protection against all nine serotypes of S. pneumoniae that were tested (130). Moreover, mouse monoclonal antibodies to pneumolysin could protect against otherwise fatal infection following intranasal inoculation of S. pneumoniae (131). Humans are known to mount an antibody response to pneumolysin as a result of natural exposure to S. pneumoniae (132, 133). The expectation that human antibodies to pneumolysin may be protective is supported by evidence that purified human antibody to pneumolysin passively protects mice from challenge with virulent pneumococci (132) and by evidence that a lack of high levels of serum antibody to pneumolysin appear to predispose patients to pneumococcal pneumonia (133). Thus, it is anticipated that immunization of humans with pneumolysoid will increase resistance to pneumococcal infection. The protection elicited by a pneumolysin-containing vaccine may be enhanced by incorporation of other protection-eliciting antigens such as PspA, immunity to which enhances complement deposition and blood clearance (73, 134, 135).

Pneumolysin has also shown promise as a carrier for the otherwise poorly immunogenic PS in experimental conjugate vaccine formulations. Immunization of mice with pneumolysoid conjugated to type 19F PS elicited a strong and boostable antibody response to both protein and PS moieties and provided infant mice with a high degree of protection against challenge with S. pneumoniae (127, 136). Similar results have been reported for conjugates of native pneumolysin with type 18C PS (137). A comparison of tetravalent pneumolysoid-PS or tetanus toxoid-PS conjugate vaccines (incorporating PS types 6B, 14, 19F, and 23F) demonstrated that pneumolysoid was at least as good a carrier protein as tetanus toxoid and, in the case of type 23F, superior (138). Clearly, such antigens have the potential to evoke a significant anti-PS response, as well as an anti-virulence-protein response, thereby conferring comprehensive protection against pneumococcal disease in humans.

Other Choline Binding Proteins

Access to the pneumococcal genome sequence facilitated the search for additional vaccine antigens, because it enabled entire families of genes encoding proteins with recognizable structural features to be targeted (139). The choline binding proteins are good examples of this approach. Although several members of this family were previously identified by conventional techniques, such as elution from the cell surface with choline, a search of the genome sequence identified a dozen or so functional genes encoding proteins with choline-binding motifs. Site-specific mutagenesis was then used to demonstrate that five of the novel choline binding proteins (CbpD, CbpE, CbpG, LytB, and LytC) were involved in in vitro adherence to epithelial cells, nasopharyngeal colonization, or sepsis, thereby identifying them as vaccine candidates (140). LytB and LytC are unusual in that their choline binding domains are located in the N-terminal part of the molecule, while the C-terminal portions have murein hydrolase activity (141). Purified recombinant LytB and LytC were subsequently tested for protective efficacy as part of another large-scale study. Immunization with these proteins conferred significant protection against intraperitoneal challenge in mice, although the degree of protection observed was marginally less than that observed using PspA, which was used as a control antigen (142).

Another choline binding protein with cell wall modification (in this case amidase) activity is the major pneumococcal autolysin LytA. Mutatgenesis of the lytA gene prevents the autolysis of pneumococci that occurs spontaneously in stationary-phase cultures, or on addition of deoxycholate and also attenuates virulence in mouse models of sepsis. It might seem paradoxical that inactivation of what is essentially a suicide gene could have such an effect. However, LytA is largely responsible for release of cell-associated pneumolysin, inflammatory cell wall degradation products, and other cell-associated virulence factors, so prevention of autolysis might be of considerable benefit to the host (116, 118, 143). Exogenous antibody to LytA is capable of penetrating the surface layers of the pneumococcus and inhibiting autolysis and release of pneumolysin in vitro. Active immunization of mice with purified LytA also elicited a similar degree of protection as pneumolysoid against challenge with fully virulent pneumococci, but it conferred no significant protection against challenge with high doses of a pneumolysin-negative strain. This suggested that the LytA-induced protection is mediated largely through blockade of pneumolysin release (144).

Lipoproteins

The pneumococcal genome includes over 30 putative lipoproteins, with prolipoprotein signal peptidase recognition sequences (LXXC) (145). This so-called lipobox motif directs covalent attachment of a diacyl glycerol moiety to the N-terminal Cys residue of the mature protein, anchoring it to the outer face of the plasma membrane. Thus, they are located beneath the cell wall and the capsule in S. pneumoniae. These lipoproteins have diverse functions, the commonest being substrate binding components of ATP binding cassette (ABC) transport systems, and many are important for growth and survival of the pneumococcus in vitro and in vivo. Their cellular location suggests that they are not exposed on the cell surface to any significant extent, which in turn suggests that they are unlikely to elicit opsonic antibodies. However, this does not necessarily preclude their utility as vaccine targets, since exogenous antibody may diffuse through the capsule and cell wall layers and inhibit the biological function of the lipoprotein. Indeed, several pneumococcal lipoproteins have been shown to have potential as vaccine antigens, as discussed below.

Pneumococcal surface antigen A (PsaA)

PsaA is a highly conserved 37-kDa lipoprotein produced by all pneumococci. It was initially thought to be an adhesin based on sequence homology with putative lipoprotein adhesins of oral streptococci, but it is actually the metal binding component of an Mn²⁺-specific ABC transport system (146). Defined psaA-negative mutants of S. pneumoniae are virtually avirulent for mice and exhibit markedly reduced adherence in vitro to human type II pneumocytes (147, 148). This is presumed to be a consequence of growth retardation due to an inability to scavenge Mn²⁺ in vivo, as well as pleiotropic effects on expression of a range of cellular processes or virulence factors. Intracellular Mn²⁺ appears to play a critical role in the regulation of expression of oxidative stress response enzymes and intracellular redox homeostasis, and psaA-negative pneumococci exhibit hypersensitivity to superoxide and hydrogen peroxide (147, 149).

One study has shown that parenteral immunization of mice with purified PsaA in the presence of strong adjuvants elicits significant protection against systemic challenge with S. pneumoniae (150). However, in other studies immunization with PsaA elicited only marginal protection and was less efficacious than pneumolysoid in an intraperitoneal challenge model (134, 151). The dimensions of PsaA (approximately 7 nm at its longest axis) (152) are such that if it is indeed anchored to the outer face of the cell membrane via its N-terminal lipid moiety, it is unlikely to be exposed on the outer surface of the pneumococcus. This is consistent with the fact that whereas the known surface-exposed domains of PspA and PspC are variable, the amino acid sequence of PsaA is highly conserved (153). Gor et al. (151) used flow cytometry to compare the surface accessibility of PsaA and PspA to exogenous specific antibodies in 12 S. pneumoniae strains. PspA was readily detectable on the surface of all strains, whereas PsaA was not. This directly correlated with the protective efficacy of either active or passive immunization with the respective protein or antibody; significant protection against systemic challenge was achieved using PspA or anti-PspA, but not using PsaA or anti-PsaA. Given the virtual absence of surface exposure, any protection elicited by immunization with PsaA is unlikely to be a consequence of enhanced opsonophagocytic clearance. Rather, it is presumably due to in vivo blockade of ion transport, which necessitates diffusion of antibody through the capsule and cell wall layers. Such penetration of antibody is likely to be concentration-dependent, and thus, high anti-PsaA titers may be required for protection. Moreover, accessibility of PsaA to exogenous antibody may be influenced by the thickness of the capsule, which may vary from strain to strain. Expression of pneumococcal capsule biosynthesis genes has also been shown to be upregulated during invasive infection (154). In contrast, pneumococci colonizing the nasopharynx are thought to downregulate capsule expression, thereby facilitating interaction between surface adhesins and the host mucosa. Consistent with this hypothesis, several studies have shown that intranasal immunization of mice with PsaA in the presence of strong mucosal adjuvants such as cholera toxin B subunit significantly reduces the level of nasopharyngeal carriage of S. pneumoniae (82, 155). A lesser but still significant reduction in susceptibility to carriage was also achieved by subcutaneous immunization of mice with synthetic lipidated multiantigenic PsaA peptides (156). Thus, at least in the nasopharynx, PsaA may be accessible to antibody. Immunization with a PsaA-cholera toxin B subunit fusion protein also significantly reduced carriage of S. pneumoniae in mice without significantly disturbing the oropharyngeal microflora (157).

Iron transporter lipoproteins PiuA and PiaA

Two other metal-binding lipoproteins have been proposed as pneumococcal vaccine antigens. These proteins, designated PiuA and PiaA, are components of two separate ABC iron transport systems. At least one of these proteins is required for optimal growth of pneumococci in iron-depleted media, and they are capable of acquiring iron from hemoglobin (158). Indeed, PiuA has been shown to be capable of directly binding both hemin and hemoglobin (159). PiuA and PiaA are produced by all pneumococci, and their genes are highly conserved (160). Mutagenesis studies have shown that both proteins contribute to virulence in mice using both lung and intraperitoneal models of infection (158). They are immunologically cross-reactive, and immunization of mice with either protein conferred a similar degree of protection against intraperitoneal challenge to that elicited by the pneumolysoid PdB. Moreover, immunization with a combination of PiuA and PiaA resulted in additive protection (161). Although a direct comparison has not been conducted, immunization with either PiuA or PiaA provided a higher degree of protection against systemic disease than that previously published for PsaA, using the same mouse model and S. pneumoniae challenge strain (134). Like PsaA, PiuA and PiaA are predicted to be attached to the outer face of the plasma membrane (159), and so the superior protective efficacy of the latter proteins ought not to be due to a difference in accessibility to exogenous antibody. However, Jomaa et al. (160) have shown by flow cytometry that both PiaA and PiuA are accessible to exogenous antibody in intact bacteria and that these antibodies stimulate in vitro opsonophagocytic activity, particularly in the presence of complement. They also reported that the antibodies did not appear to interfere with iron uptake in vitro. Mucosal immunization with PiuA and PiaA has also been shown to elicit antibody responses both in serum and respiratory secretions, which protected mice against intranasal challenge (162). The reason for the apparent difference in surface accessibility between PsaA and the two iron-binding lipoproteins is unclear, given their predicted location. One possibility is that in the latter two cases, at least a proportion of the proteins are released from the membrane and are then able to bind to more exposed domains on the pneumococcal surface, where they can interact with exogenous antibody more freely. Regardless of the underlying mechanism, available data suggest that PiaA and PiuA have more promise than PsaA as vaccine antigens, as least for prevention of systemic disease.

Pneumococcal Histidine Triad (Pht) Proteins

The pneumococcal histidine triad proteins are a recently recognized family of surface proteins that have an unusual polyhistidine motif, HXXHXH, repeated five or six times in their amino acid sequences. The prototype, PhtA, was discovered as part of a genome-wide screen for potential vaccine antigens (142). Over 100 proteins were expressed and tested for efficacy in a mouse model, and PhtA was 1 of only 5 that were protective. The others were the choline binding proteins LytB and LytC (discussed previously), a cell wall-associated serine protease PrtA, and another protein of unknown function, designated PvaA. Further examination of the pneumococcal genome sequence revealed three additional related open reading frames (designated PhtB, PhtD, and PhtE), each with five or six copies of the histidine triad motif. The four proteins range in size from 91 to 114 kDa and are closely related at the amino acid sequence level, exhibiting 32 to 87% identity; this similarity is strongest in the N-terminal regions (163). Although their signal peptides all contain an LXXC motif, this does not appear to function as a true lipobox, because they are not labeled by [³H]-palmitate (164). Flow cytometric analyses have shown that the C-terminal regions are more readily accessible to exogenous antibodies (163, 164), suggesting that they are tethered via their N-termini. Analysis of deletion and point mutants subsequently identified a three-amino-acid region in PhtD (Q27-H28-R29) that is critical for surface attachment (165). The histidine triads are believed to form a novel Zn²⁺ binding motif (166), and the pht genes are regulated by the Zn²⁺-dependent repressor AdcR (167). A major function of Pht proteins is to facilitate Zn²⁺ import via AdcAII, one of two Zn²⁺ binding lipoproteins associated with the AdcBC permease (168). Of the five His triad motifs in PhtD, that closest to the N-terminus appears to have the greatest impact on Zn²⁺ recruitment (169). Pht proteins may also inhibit complement deposition on the pneumococcal surface through binding of factor H (167). They exhibit substantial functional redundancy; significant effects on virulence in mouse models of pneumococcal sepsis, on AdcAII-dependent Zn²⁺ uptake, and on complement deposition in vitro were only observed when all four pht genes were deleted (167, 168).

There is a high degree of protein sequence conservation among individual Pht proteins from diverse S. pneumoniae serotypes (164), which combined with a degree of immunological cross-reactivity between the proteins augers well for broad strain coverage for these candidate vaccine antigens. Immunization with purified PhtA, PhtB, or PhtD has been shown to confer significant protection against intraperitoneal challenge with type 3, 6A, and 6B and one of two type 4 S. pneumoniae strains (142, 163). PhtD has also been shown to protect against intranasal challenge with a type 3 strain (170), while immunization with either PhtB or PhtE also protects against type 3 pneumococci in models of sepsis and pneumonia (164). In this latter study, immunization experiments with truncated PhtE fragments localized the protective epitopes to the more surface-exposed C-terminal region of the molecule. However, notwithstanding these promising results, the only direct comparative studies of the protective efficacy of Pht proteins with other well-characterized antigens indicate that the level of protection elicited is no better than that achieved by either PspA or pneumolysoid (142, 171).

Sortase-Dependent Surface Proteins

Sortase-dependent surface proteins of Gram-positive bacteria are identifiable by the presence of a C-terminal anchoring motif, which consists of a conserved LPXTG sequence, followed by a hydrophobic domain and usually a tail of positively charged residues. This motif is recognized by sortase, a membrane-localized cysteine protease which cleaves between the T and G residues and covalently links the processed protein to the peptidoglycan cross-bridges (172). In pneumococci, inactivation of the sortase gene releases known sortase-dependent surface proteins, such as the major pneumococcal neuraminidase NanA, and reduces adherence to pharyngeal cells (173). NanA-deficient mutants of S. pneumoniae have been shown to have a reduced capacity to colonize the upper and lower respiratory tracts of mice (174, 175) and the nasopharynx and middle ear cleft of chinchillas (176). An early study indicated that purified NanA had modest but significant protective efficacy (weaker than pneumolysin) in a mouse sepsis model (144). It was also shown to be protective against both carriage and otitis media in chinchillas (177, 178).

Significant attention has also been paid to pilus-like structures on the surface of S. pneumoniae, which contribute to virulence (179). These are encoded on the rlrA pathogenicity islet, which is present in some but not all clinical isolates. The rlrA islet encodes a transcriptional regulator, three pilus structural components (RrgA, RrgB, and RrgC), and three sortases (SrtB, SrtC, and SrtD) which are required for pilus assembly (179, 180). RrgB is the major pilin, while RrgA and RrgC are ancillary pilin subunits decorating the shaft and tip (181). Immunization of mice with purified RrgA, RrgB, or the combination of all three pilus proteins elicited significant protection against intraperitoneal challenge with the S. pneumoniae strain from which the proteins originated (182). However, the utility of pilus proteins as stand-alone vaccine antigens is limited by the fact that there are three sequence clades of the rlrA islet, each associated with distinct S. pneumoniae clonal groups. At the amino acid sequence level, RrgB is the most divergent, with only about 50% identity between clades (183), and the extent of cross-protection is unknown. Moreover, rlrA-positive pneumococci account for only about 30% of strains, and the majority of these belong to serotypes covered by PCV7 (183, 184). Interestingly, a second pneumococcal pilus locus has recently been described, and this was present in about 16% of strains, again belonging to discrete clonal groups, all but one of which lacked rlrA (185). However, the vaccine potential of these distinct pilus proteins is uncertain.

Other sortase-dependent pneumococcal surface proteins have been proposed as vaccine candidates, including hyaluronidase (Hyl) and the IgA1 protease (Iga). The latter is of interest because its sortase motif is located in the N-terminal region of the molecule, but it is nevertheless essential for proper function and surface localization (186). However, although these proteins have been shown to contribute to pathogenesis using either in vitro or in vivo models, Hyl is at best a weak protective antigen (187), while Iga is yet to be tested individually for protective efficacy.

Other Protein Vaccine Candidates

A number of other apparently surface-associated pneumococcal proteins with at least theoretical vaccine potential have been identified, in most cases using immuno-proteomic approaches. This list includes proteins that lack export or anchorage signals and would have been predicted to be cytoplasmic and hence dismissed by previous motif-based searches. Examples include metabolic enzymes such as enolase (which also binds plasminogen) (188), 6-phosphogluconate dehydrogenase (a putative adhesin) (189), fructose-biphosphate aldolase, and glyceraldehyde-3-phosphate dehydrogenase (190), as well as the heat-shock protease ClpP (191). Another candidate is the putative proteinase maturation protein PpmA (192), although its degree of surface exposure and protective efficacy has been questioned (151).

Giefing et al. (193) conducted a comprehensive study of the antibody repertoire induced in humans during natural pneumococcal infection by extensive screening of E. coli display libraries representing the entire S. pneumoniae proteome with carefully selected panels of convalescent patient sera. Sequence analysis of reactive clone inserts identified not only proteins that reacted consistently with the sera, but also the key epitopes therein. Over 140 immunoreactive proteins were identified, and many of the vaccine candidates referred to above, most notably PspA, PspC, PhtA/B/D/E, NanA, Iga, LytC, and LytA, were detected with high frequency. Interestingly, about 20% of the proteins represented unknown gene products and thus may have been missed in previous motif-based in silico screens. A subset of the identified proteins was selected on the basis of further seroreactivity screens of synthetic peptides, surface localization, and opsonophagocytic studies using epitope-specific hyperimmune mouse sera and PCR-based gene distribution analyses. Twenty selected proteins were then purified and tested for protective immunogenicity in mouse models of sepsis and pneumonia. Significantly, two proteins (PcsB and StkP) provided a degree of protection similar to that seen for PspA. Both are highly conserved and appear to play an important role in cell separation and/or formation of division septa. PcsB and StkP both protected against a variety of challenge strains and were superior to PspA when the challenge strain expressed a dissimilar family/clade of PspA (193).

A recent study evaluated the vaccine potential of 52 pneumococcal proteins selected on the basis of genetic conservation across disease-causing serotypes and bioinformatic prediction of antibody binding to the target antigen (194). Seven proteins induced protective responses in a murine infection model, namely, the Pht family proteins PhtB, PhtD, and PhtE and the sortase family proteins PrtA, NanA, PavB, and Eng.

From the above it can be seen that there are a number of pneumococcal proteins that exhibit potential as vaccine antigens. However, assessment of their protective efficacy has generally been carried out in different laboratories using a variety of animal models and challenge strains. Relatively few direct comparative protection studies have been performed to determine which of these proteins provides the strongest protection against the widest variety of S. pneumoniae strains.

Combination Protein Vaccines

The vast majority of the pneumococcal proteins under consideration as vaccine antigens are directly or indirectly involved in the pathogenesis of pneumococcal disease. Mutagenesis of some combinations of virulence factor genes, for example, those encoding pneumolysin and either PspA or PspC, PspA, or PspC or all three genes, has been shown to synergistically attenuate pneumococcal virulence in animal models, suggesting that the respective proteins function independently in the pathogenic process (195, 196). This strongly suggests that immunization with combinations of these antigens might provide additive protection. Moreover, there may be differences in the relative protective capacities of the individual antigens against particular S. pneumoniae strains, particularly for surface-exposed antigens that exhibit some degree of sequence variation. Thus, a combined pneumococcal protein vaccine may elicit a higher degree of protection against a wider variety of strains than any single antigen.

Immunization of mice with a combination of the pneumolysoid PdB and PspA provided significantly increased protection against intraperitoneal challenge than immunization with either protein alone. However, combining either protein with PsaA did not result in enhanced protection (134). The potential benefits of combination protein vaccines are also well illustrated using a mouse model of nonbacteremic pneumonia, which closely reflects the commonest form of pneumococcal respiratory disease in humans (135). In this system, subcutaneous immunization (using alum adjuvant) with either PdB or PspA, but not PsaA, significantly reduced numbers of S. pneumoniae in the lungs 7 days after challenge. A significant additional reduction in bacterial load was achieved by immunization with a combination of PdB and PspA, but not when either protein was combined with PsaA (135). These findings contrast with those obtained using mucosal (intranasal) immunization with the same proteins with cholera toxin B subunit as the adjuvant. As discussed previously, immunization with either PspA or PsaA, but not PdB, reduced the level of carriage of S. pneumoniae after intranasal challenge, and the combination of PspA and PsaA was more effective than either antigen alone (82). These findings suggest that PsaA is either more important for survival of S. pneumoniae in the nasopharynx than in the lung or that it is more accessible to exogenous antibody in the former niche. On the other hand, pneumolysin appears to play only a minor role during the colonization phase but is clearly important once the organism has been aspirated into the lungs. Thus, optimum vaccine formulation will be dependent upon the mode of vaccine delivery and the stage of the pathogenic process being targeted for immunoprophylaxis. In a more recent study, a variety of multivalent protein-based vaccines comprised various combinations of full-length or peptide regions of the immunogens pneumolysin, PspC/CbpA, or PspA. Optimum protection against multiple challenge strains in models of sepsis, pneumonia, and meningitis was obtained using a combination comprising pneumolysin toxoid with an L460D substitution, with two protective peptide epitopes from PspC/CbpA (known to be involved in binding to the host polymeric immunoglobulin receptor and the laminin receptor, respectively) fused to the N- and C-termini, as well as a polypeptide derived from the conserved proline-rich region and SM1 regions of PspA (197).

Only a handful of other pneumococcal protein combinations have been tested for additive protective immunogenicity. Immunization with both the iron transporters PiuA and PiaA was more effective than either antigen alone (161). A combination of pneumolysoid (PdB), PspA, and PspC has also shown stronger protection than single or paired antigens (171). In that study, combinations of PhtB and PhtE were less protective than those involving two or more of PdB, PspA, and PspC, and inclusion of PhtB did not improve the protection imparted by either PdB/PspA or PdB/PspC (171). Additive protection relative to individual components has also been demonstrated for a trivalent formulation comprising another pneumolysin toxoid (PlyD1), PhtD, and pneumococcal choline binding protein A (PcpA) in an infant murine otitis media model (198). Clearly, additional comparative studies of the protective efficacy of the better-characterized proteins, as well as the more recently identified vaccine candidates (both singly and in combination), are desirable to enable informed decisions on the optimum formulation of a protein-based pneumococcal vaccine. However, further development of some potentially efficacious combinations has been frustrated by the intellectual property landscape surrounding individual components, which impacts freedom to operate for commercial partners taking experimental vaccine formulations forward into the clinic.

Of the pneumococcal combination protein vaccines that have reached phase II human trials, the PlyD1/PhtD/PcbA formulation referred to above has been shown to be safe, well-tolerated, and immunogenic in adults, toddlers, and infants (199). A formulation comprising the currently licensed 10-valent PS conjugate vaccine, supplemented with chemically detoxified pneumolysin (dPly) and PhtD was also safe and immunogenic in infants (200, 201), but this vaccine formulation did not reduce nasopharyngeal carriage of nonconjugate vaccine serotypes (202). A cell-free protein extract from a derivative of S. pneumoniae TIGR4 expressing a nontoxic pneumolysin gene, enriched for various conserved surface proteins by anion exchange, was also safe and immunogenic in adults at phase I (203).

PNEUMOCOCCAL WHOLE-CELL VACCINES (WCV)

An alternative strategy for eliciting non-serotype-dependent protection is immunization with a simple vaccine comprising killed nonencapsulated pneumococci that display critical protein antigens on their surface. A study looking at historical data from the 1918 influenza pandemic showed that WCVs had the ability to confer cross-protection against different pneumococcal serotypes and that such vaccines were protective against influenza-associated pneumonia (204). A WCV comprising a chemically killed lytA-deletion mutant of S. pneumoniae Rx1 expressing nontoxic pneumolysin has also been shown to elicit protection against sepsis and nasopharyngeal carriage in mouse models. Protection against sepsis was dependent on antibodies to noncapsular antigens, while prevention of carriage required CD4+ T cell-dependent responses (205, 206). The WCV approach has potential cost advantages over the currently licensed conjugate vaccines and the various multicomponent purified protein formulations currently in clinical development. A chemically killed WCV formulation was safe and immunogenic in a phase I trial and is in further clinical development (207).

Another study has used γ-irradiation to inactivate a similar unencapsulated S. pneumoniae strain. Intranasal vaccination with this WCV (without adjuvant) elicited serotype-independent protection in lethal challenge models of pneumococcal pneumonia and sepsis. Vaccine efficacy was shown to be reliant on B-cells and interleukin-17A responses. Interestingly, immunization promoted interleukin-17 production by innate γδ T cells, not T helper 17 (Th17) cells (208). Further enhancement of immunogenicity was observed when this vaccine was coadministered with a γ-irradiated influenza virus vaccine that also provides heterotypic protection against the virus (209).

MUCOSAL IMMUNITY

The development of vaccines that can elicit mucosal immunity could be very important because (i) mucosal immunity is thought to provide the best protection against carriage (61, 210), (ii) nasopharyngeal carriage is the human reservoir for pneumococci, and herd immunity depends on immunity to carriage (39, 82, 210), (iii) at least a short duration of carriage is required prior to almost all invasive disease (211), and (iv) epidemiologic data suggest that meningitis frequently occurs with strains of capsular types which seldom cause sepsis but which are readily carried (50, 212, 213). In the section above about PspA and PspC, mucosal immunity studies are described that used a cationic cholesteryl group-bearing pullulan nanogel, which greatly enhanced mucosal immunity in mice and, like many other recent enhancers of mucosal immunity (214, 215), is not transported to the brain (83, 84).

CONCLUSIONS

The ongoing high global morbidity and mortality associated with pneumococcal disease and the complications caused by increasing rates of resistance to antimicrobials have underpinned extensive efforts in recent years to develop more effective vaccination strategies against S. pneumoniae. These efforts have benefited from a better understanding of the mechanisms of pathogenesis of pneumococcal disease and the advances made possible by the advent of recombinant DNA technology and access to genome sequence data. The polyvalent PS vaccines have doubtless prevented a large number of deaths from invasive disease in recipients belonging to those patient groups for whom this vaccine is currently recommended. The newer PS-protein conjugate formulations confer a very high degree of protection in young children against the included serotypes and may also have an impact on the prevalence of drug-resistant strains. However, there is now general acceptance that this vaccination approach is not without its drawbacks, and as explained above, the initially substantial clinical benefits that are expected to be derived from widespread use of conjugate vaccines may diminish with time. It will take many years for the overall impact of conjugate vaccines on disease burden and the population biology of S. pneumoniae to become apparent. At the very least, use of the conjugate vaccines will buy time for development of less expensive, largely non-serotype-specific vaccines based on combinations of protein antigens. It must be emphasized, however, that the success of these protein vaccines is not dependent upon real or perceived failure of the conjugates. Rather, the two approaches should be viewed as complementary, each having an important role to play in global prevention of pneumococcal disease. Nor should development of parenteral protein vaccines impede future research on mucosal- or DNA-based delivery systems, which may further improve presentation of protective antigens to the immune system, thereby optimizing host responses and herd immunity.