|Home | About | Journals | Submit | Contact Us | Français|
Much of the efficacy of current pneumococcal conjugate vaccines lies in their ability to decrease carriage of vaccine serotypes in the population. Novel and more-broadly acting vaccines would also need to target carriage in order to be as effective. We have previously shown that model murine carriage of Streptococcus pneumoniae can elicit antibody-dependent immunity and can protect against a virulent heterologous challenge strain. This study set out to identify S. pneumoniae surface antigens that may elicit cross-reactive antibodies following colonization. Western blot analysis using sera from colonized mice identified the previously characterized immunogens pneumococcal surface protein A (PspA), putative proteinase maturation protein A (PpmA), and pneumococcal surface adhesin A (PsaA) as such antigens. Using flow cytometry, PspA was found to be the major target of surface-bound cross-reactive IgG in sera from TIGR4Δcps-colonized mice, with a modest contribution from PpmA and none from PsaA. In human sera, however, only mutants lacking PpmA were shown to have reduced binding of surface IgG compared to wild-type strains, suggesting that prior exposure to S. pneumoniae in humans may induce PpmA antibodies. We also investigated if cross-reactive antibodies induced by these antigens may be cross-protective against carriage. Despite the immunogenicity of PspA, PpmA, and PsaA, mice were still protected following colonization with mutants lacking these antigens, suggesting they are not necessary for cross-protection induced by carriage. Our findings suggest that a whole-organism approach may be needed to broadly diminish carriage.
Streptococcus pneumoniae (the pneumococcus) is a major human pathogen responsible for over 1 million deaths annually worldwide. The pneumococcus is a leading cause of common mucosal infections, including otitis media and pneumonia, as well as disseminated diseases, such as sepsis and meningitis. Treatment is complicated by the increasing prevalence of β-lactam resistance and by strains resistant to multiple classes of antibiotics. This has highlighted the need for preventative strategies against the spectrum of pneumococcal diseases.
The advent of the pneumococcal conjugate vaccine (PCV7) has led to reductions of pneumococcal disease in children and adults (45, 47), by direct vaccination and through herd immunity, respectively. Despite the success of this vaccine in reducing invasive pneumococcal disease (IPD), the level of protection from mucosal infections is more limited (14, 15). One of the major issues with PCV7 is that it targets the serotype-determining polysaccharide capsule. Although the capsule is an important virulence factor and a potent antigen when conjugated to a protein carrier, antibodies generated are thought to only protect against a homologous capsule type. There are at least 91 distinct pneumococcal capsule types, and although isolates of the seven serotypes included in the current vaccine are responsible for 80% of IPD in the United States, vaccination with capsular polysaccharides of a limited number of types has led to an increase in the prevalence of serotypes not included in the vaccine (serotype replacement). In addition, the distribution of serotypes responsible for IPD varies by location; therefore, vaccines need to be tailored to each geographic region to ensure the greatest level of protection. This geographic specificity, coupled with the complexity of the vaccine, contributes to the prohibitive cost for those in most need in the developing world. An inexpensive broad-spectrum vaccine against a common antigen(s) could overcome the limitations of PCV7.
Pneumococcal antigens that are common to all or most serotypes have received much interest as vaccine targets for their potential to induce broad protection. Some of these include surface proteins (choline binding proteins [8, 9], lipoproteins [6, 40], a toxin , histidine triad proteins , and sortase-dependent surface proteins) and cell wall structural components (16, 27, 43; for a review, see reference 41). These antigens given alone or in combination elicit systemic and/or mucosal protection when administered by a variety of methods with adjuvants in animal models. Some of these protein antigens have been confirmed by unbiased genomic approaches, looking for antigens recognized by antibodies from patients convalescing from pneumococcal diseases (16, 48). The success of studies involving these antigens highlights the potential for common surface proteins in protecting against IPD.
The human nasopharynx is the site of asymptomatic colonization, the organism's carrier state, and is also the source of horizontal transfer. Colonization is also considered a prerequisite to disease (5). Young children, the main reservoir of the pneumococcus, are heavily colonized by S. pneumoniae, and many acquire one or more strains sequentially or simultaneously. Colonization rates decline significantly as age increases, suggesting that this early colonization may be an immunizing event (19). However, the immune mechanism responsible for the decline in colonization has yet to be fully defined. It is clear that reducing colonization prevents pneumococcal disease. For example, experience with PCV7 has demonstrated that decreased colonization in vaccine recipients reduces transmission and leads to decreased disease in those who have not been vaccinated (33, 47). This suggests that the effectiveness of any novel vaccine that decreases carriage would be magnified in the community because of the contribution of herd immunity.
The focus of this study was to take an unbiased look at which pneumococcal antigen(s) induces broadly cross-reactive antibodies following nasal colonization. We have previously shown that carriage of S. pneumoniae (live attenuated vaccine) can elicit antibody-dependent immunity and can also protect against a heterologous challenge strain (39). Here, we use this approach as a tool to identify cross-reactive antigens, by dissecting out the main targets of the humoral immune response using a mouse model of nasal colonization.
S. pneumoniae strains were grown in tryptic soy broth (BD, Franklin Lakes, NJ) at 37°C in a nonshaking water bath. Strains used in this study were selected because of their ability to efficiently colonize the murine nasopharynx and included 6A (type 6A, mouse virulent clinical isolate) (23), TIGR4 (type 4 clinical isolate, genome sequence strain) (44), and 23F (type 23F strain previously used for human studies) (29) (Table (Table1).1). Unencapsulated (Δcps) TIGR4 and 6A strains can also colonize the mouse nasopharynx for up to 2 weeks (34). 6A and TIGR4 have previously been shown by multilocus sequencing to be different clonal complexes (39). The pspA gene from each strain has been sequenced. TIGR4 expresses PspA from family 2 (clade 3), whereas both 6A and 23F express PspAs from family 1 (clades 2 and 1, respectively). All strains were passaged intranasally in mice prior to preparation of frozen stocks.
To generate an unencapsulated mutant, the cps operon was interrupted in 6A and TIGR4 using the Janus cassette (34). The 6AΔpspA mutant was constructed by insertion deletion as previously described (39). The mutation was transferred to TIGR4 and 23F using sequential back-transformation with lysates of mutant 6AΔpspA, selecting for spectinomycin resistance (200 μg/ml) (49).
The ΔppmA and ΔpsaA mutants were also constructed by insertion-deletion, with an erythromycin (erm) cassette, from shuttle vector pMU1328 (1), replacing the majority of the targeted genes. The ΔppmA mutant was constructed using overlap extension PCR (21). Primer pairs PpmAF7/PpmAOER1 and PpmAOEF2/PpmAR6 were used to amplify the two flanking regions of ppmA from TIGR4, deleting most of the gene and adding nucleotides complementary to the erm cassette. Primers PpmAOEF1 and PpmAOER2 were used to amplify the erm cassette, adding nucleotides complementary to the ppmA gene flanking regions at the beginning and end of the sequences. The three fragments were pieced together using the primers PpmAF8 and PpmAR7 by overlap extension PCR. This PCR construct, with the erm cassette replacing most of the ppmA gene, was transformed into TIGR4, 6A, and 23F with selection for erythromycin resistance (1 μg/ml).
The ΔpsaA mutant was constructed by amplifying the gene using primers PsaAFHind32 and PsaAR. The PCR product was cloned into the TOPO PCR2.1 plasmid and transformed into One Shot Top10 chemically competent Escherichia coli using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA). A deletion was made in this gene using inverse PCR with primers InvpsaAFmef1 and InvpsaARmef1, and the erm cassette was inserted into the plasmid cut with MfeI. S. pneumoniae ΔpsaA mutants were generated by transformation with the plasmid DNA, with selection for erythromycin resistance (1 μg/ml). All mutants were confirmed by PCR and/or Western blotting. Primer sequences can be found in Table Table22.
Six-week-old female C57BL/6 (wild type), B6.129-S2-Igh-6tm1Cgn/J (μMT; Jackson Laboratories, Bar Harbor, ME), and IgA−/− (kind gift from Yimin Yu, University of Pennsylvania) mice were housed in accordance with Institutional Animal Care and Use Committee protocols. μMT mice contain a targeted mutation in the heavy chain locus of IgM and do not produce mature B cells or antibody (24). IgA−/− mice contain a targeted deletion in the alpha-chain constant region and do not express IgA (20). All mice are in the C57BL/6 background. Mice were colonized using a previously described model of nasopharyngeal colonization with S. pneumoniae (30). Briefly, mice were inoculated intranasally (i.n.) without anesthesia with 10 μl containing 1 × 107 to 5 × 107 CFU of phosphate-buffered saline (PBS)-washed, mid-log-phase S. pneumoniae cells applied atraumatically to the nares. At the times indicated, the animals were sacrificed, the tracheas cannulated, and 200 μl of PBS instilled. Lavage fluid was collected from the nares for determination of viable counts of bacteria in serial dilutions plated on selective medium containing neomycin (5 μg/ml for TIGR4 and 20 μg/ml for 6A) to inhibit the growth of contaminants. The lower limit of detection for bacteria in lavage culture was 20 CFU/ml.
Mice received a one- or two-dose intranasal immunization (where noted) with 10 μl containing 107 CFU without adjuvant, with the second dose given 2 weeks after the first dose. Mice were challenged intranasally with 10 μl containing 1 × 107 to 5 × 107 CFU of S. pneumoniae strains 5 weeks after the final immunization. We have previously shown that Δcps strains are attenuated and are cleared from the upper respiratory tract by 2 weeks postinoculation (34). Mice were observed for signs of sepsis, such as decreased activity, grooming, eating, or drinking, over an 8-day period postchallenge. Animals showing signs of sepsis were euthanized, and the spleens were cultured to confirm the presence of pneumococci. After 8 days, the remaining animals were euthanized and nasal washes were obtained for quantitative culture. The first colonizing strain was never detected by selective plating of nasal lavages, including in those obtained from immunodeficient mice. Blood was also collected from cardiac punctures, and serum was stored at −20°C for flow cytometric analysis.
Whole-cell lysates were prepared as follows: bacteria were grown to mid-log phase in tryptic soy broth, and 1-ml aliquots were pelleted, washed, and resuspended in 50 μl of Laemmli sample buffer (25) and boiled for 5 min. Twelve-microliter aliquots of samples were loaded onto 10% polyacrylamide gels (Bio-Rad, Hercules, CA).
Fractionation of whole bacteria was performed by growing large cultures of bacteria to mid-log phase, washing, and then resuspending in fractionation buffer (30 mM Tris, 150 mM NaCl). Cultures were then passed through a French press system (American Instrument Co., Silver Spring, MD) twice to lyse the bacteria. Unlysed bacteria were removed by centrifugation at low speed (7,500 × g). The lysed bacteria were separated into soluble and insoluble fractions by ultracentrifugation at 275,000 × g, with the soluble fraction remaining in the supernatant fluid. Insoluble fractions were washed twice with PBS prior to resuspension in 1 ml of PBS. The purity of fractions was confirmed by Western blotting using antibodies against known membrane protein PsaA and cell wall component lipotechoic acid (data not shown). Samples were stored at −20°C for use in gel electrophoresis.
Choline washing of insoluble fractions was performed by incubating 1 part protein fraction with 1 part choline chloride (Sigma, St. Louis, MO) (4% in PBS) to a final concentration of 2% choline. Samples were rotated at 4°C for 1 h. Choline binding proteins (CBPs) were then separated from fractions by centrifugation, with the CBPs remaining in the supernatant fluid (50). Samples were stored at −20°C for use in gel electrophoresis.
One-dimensional SDS-PAGE was performed using the Mini-Protean II system (Bio-Rad). Protein samples were suspended in Laemmli sample buffer and boiled for 5 min prior to electrophoresis at 100 V. Two-dimensional SDS-PAGE involved separation of proteins based on isoelectric point and by molecular weight, respectively. Isoelectric focusing (pI 4.7 to 5.9) was carried out in a Protean IEF cell (Bio-Rad), using 7-mm ReadyStrips, according to the manufacturer's instructions. The proteins were then separated in the second dimension by using 10% polyacrylamide gels in a Mini-Protean II system. The proteins were stained in the gels using Coomassie brilliant blue R-250 (Fisher Scientific, Pittsburgh, PA).
Protein mixtures were separated by one-dimensional and two-dimensional SDS-PAGE and transferred to a polyvinylidene diflouride (PVDF) membrane (Thermo Scientific, Waltham, MA) by using the Trans-Blot SD semidry transfer system (Bio-Rad) at 18 V. Gels from two-dimensional SDS-PAGE were half transferred (18 V for 0.3 h, compared to 0.6 h for one-dimensional SDS-PAGE) and following transfer the remaining gel was stained using Coomassie brilliant blue R-250 to obtain a stained gel and Western blot membrane pair. Membranes were then blocked in PBS supplemented with 1% bovine serum albumin prior to incubation with mouse serum (pooled from five mice) at a 1:1,000 dilution at room temperature overnight. Bound antibody was detected by using anti-mouse secondary antibody conjugated to alkaline phosphatase (Sigma) and 5-bromo-4-chloro-3-indolylphosphate-nitroblue tetrazolium (Fisher) development.
Spots identified by Western blotting as cross-reactive were traced on the corresponding Coomassie-stained gel, and protein spots of interest were excised from the gel using a pipette tip. The protein within the gel plug was trypsin digested and injected onto a high-performance liquid chromatography (HPLC) C18 column to separate the digested peptides. The separated peptides were sprayed into an LTQ ion-trap mass spectrometer (Thermo Scientific). Mascot software (Bloomington, IN) was used to search bacterial databases for sequence similarities. Cutoffs were assigned as a protein score of >70 with a unique peptide value of >2.
One hundred microliters of mid-log-phase bacteria was pelleted and washed in Hanks buffer (Invitrogen) supplemented with 5% fetal calf serum (HFCS; Thermo Scientific). Cells were then incubated with antibody source (1% human sera or 10% pooled mouse serum), diluted in HFCS, for 45 min at 37°C. Following washing in HFCS, the cells were incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody against IgG (Sigma) for 30 min on ice at a concentration of 1:100 in HFCS. After washing, the cells were resuspended in 200 μl of 1% formaldehyde. Flow cytometric analysis was conducted on a FACSCalibur machine with CellQuest software (BD, Franklin Lakes, NJ). Bacteria were gated, and 20,000 events were collected. The quantity of antibody bound to bacterial cells was calculated by measuring the geometric mean fluorescence intensity (MFI) of the strains incubated with antibody source minus the no-antibody control, using FlowJo software (Tree Star, Ashland, OR). Human sera were obtained from healthy adults aged 23 to 50 years.
Colonization density was expressed as the log10 CFU/ml for calculation of means ± standard errors of the means. Statistical comparisons of survival and colonization among groups were made by a nonparametric test as indicated. Statistical analyses were performed using Prism 5 (GraphPad, La Jolla, CA).
Antibody was previously shown in our mouse model to be required for protection elicited by prior colonization (39). Following colonization with TIGR4Δcps, wild-type mice, but not μMT mice, which lack specific antibody, were protected from sepsis after intranasal challenge with 6A, a strain that causes a high rate of invasive infection following intranasal inoculation (Fig. (Fig.1A).1A). This demonstrates that specific antibody is particularly important in cross-protection against invasive infection from the mucosal surface. We also investigated if IgA mediates mucosal protection, because IgA is the most abundant immunoglobulin on the mucosal surface of the nasopharynx. This was carried out by colonizing mice with the 6AΔcps mutant, as unencapsulated strains are unable to cause invasive infection, followed by challenge with virulent parental 6A and monitoring survival. Unlike μMT mice, IgA−/− mice were protected during secondary challenge, suggesting that IgA is not required to prevent sepsis following intranasal challenge (Fig. (Fig.1B).1B). The role of IgA was also investigated in mucosal protection by monitoring the level of colonization of the challenge strain at day 8 postchallenge. There was a wide range of colonization densities in previously colonized IgA−/− mice; however, 40% of these mice were still able to clear colonization, suggesting that IgA is not required for mucosal protection induced by colonization (Fig. (Fig.1C).1C). Since antibody, but not IgA, is important for protection, we focused on IgG and IgM responses from colonized mice to identify bacterial antigens inducing cross-reactive antibodies.
To identify the targets of cross-reactive antibodies induced by carriage, mice were colonized with isolates of three different serotypes (4, 6A, or 23F), and the cross-reactive serum Ig response was analyzed by Western blotting. Whole-cell pneumococcal lysates were separated by SDS-PAGE and transferred to PVDF membranes, which were then probed with sera from mice colonized with one of two heterologous strains to look at cross-reactive responses. A similar banding pattern of cross-reactivity was observed in binding of serum IgM and IgG (differentiated using specific secondary anti-mouse Ig antibodies) (data not shown), and so we focused our analysis on the IgG response. When the 6A lysate was probed with sera from mice previously colonized with TIGR4Δcps, three prominent areas of cross-reactive bands were seen (Fig. (Fig.2A,2A, panel 1). Fractionation of lysates indicated bands in area 2 were in the soluble or cytoplasmic fraction (Fig. (Fig.2A,2A, panel 2). Analysis of a Δply mutant confirmed that the band in area 2 was not the previously characterized vaccine candidate pneumolysin, which may be predominantly cytoplasmic (data not shown). The cytoplasmic antigen in area 2 was excluded from further analysis, since our focus was on cross-reactive surface antigens. The antigens in areas 1 and 3 were confirmed as proteins, because prior treatment of lysates with proteinase K resulted in the complete loss of cross-reactivity (data not shown).
S. pneumoniae possesses a family of proteins on its surface that are noncovalently attached to choline residues on teichoic acid (TA). To determine if pneumococcal CBPs are recognized by cross-reactive antibodies induced by colonization, insoluble fractions were washed with 2% choline to remove CBPs. The choline wash of 6A contained a single protein of ~75 kDa that reacted with sera from mice colonized with TIGR4Δcps (Fig. (Fig.2B).2B). Bands in area 1 were identified as PspA based on the absence of these bands in a 6AΔpspA mutant. Two-dimensional (2D) electrophoresis and mass spectrometry were carried out to identify the predominant cross-reactive low-molecular-weight band (area 3) (data not shown). Peptide sequencing identified a single protein, PpmA, which was confirmed by the loss of the band in a 6AΔppmA mutant (Fig. (Fig.2C2C).
When sera from mice colonized with 6AΔcps were probed for TIGR4 whole-cell lysates, a different pattern of cross-reactivity was noted, with only a single prominent band. This band contained a single protein which was identified by 2D electrophoresis and mass spectroscopy as PsaA (data not shown). This was confirmed by the loss of the band in a TIGR4ΔpsaA mutant (Fig. (Fig.2D2D).
Mice colonized with a third isolate, 23F, induced cross-reactive antibodies that recognized PpmA in whole-cell lysates of both TIGR4 and 6A (Fig. (Fig.2E).2E). However, neither PspA nor PsaA was recognized (data not shown). Therefore, PspA, PpmA, and PsaA induce cross-reactive antibodies following murine pneumococcal colonization. However, the cross-reactivity of each antigen varies depending on the colonizing strain.
PpmA and PsaA are both highly conserved among S. pneumoniae strains (18). To determine the extent of cross-reactivity of induced antibodies, whole-cell lysates from a range of pneumococcal strains, including representatives of common serotypes, were separated by SDS-PAGE and probed with sera from mice colonized with TIGR4Δcps, 6AΔcps, or 23F. A similar banding pattern of cross-reactivity was seen with sera from TIGR4Δcps-colonized mice (Fig. (Fig.3A).3A). This serum recognized PpmA in six of six heterologous strains analyzed. Similarly, sera from mice colonized with 6AΔcps reacted with PsaA in six of six heterologous strains analyzed (Fig. (Fig.3B).3B). Sera from mice colonized with 23F also reacted with PpmA in six of six heterologous strains analyzed (Fig. (Fig.3C).3C). In sera from TIGR4Δcps-colonized mice the ~50-kDa cytoplasmic protein also appeared to be broadly cross-reactive in all strains tested. This confirmed that these antigens have the ability to generate broadly cross-reactive antibodies.
To determine the relative contribution of each antigen to the humoral immune response, flow cytometry was performed on wild-type strains and mutants lacking each of the three identified cross-reactive antigens (PspA, PpmA, and PsaA). Whole bacteria were incubated with sera pooled from five mice colonized with a heterologous strain to look at cross-reactive binding. The amount of cross-reactive surface antibody against each antigen was calculated as the difference in binding between the wild type and mutants lacking the antigen. IgG induced by colonization with TIGR4Δcps recognized the wild-type 6A, and this binding was greatly reduced in a 6AΔpspA mutant (Fig. (Fig.4A).4A). This suggests that PspA is responsible for the majority of the 6A cross-reactive surface antibody generated by TIGR4Δcps colonization. Loss of PpmA resulted in a minimal reduction in surface-bound antibody, suggesting that PpmA is the target for limited cross-reactive antibody to the bacterial surface (Fig. (Fig.4B).4B). However, the loss of PsaA had no effect on surface-bound antibody, suggesting that PsaA does not contribute to cross-reactive surface antibody (Fig. (Fig.4C).4C). This was confirmed by the lack of surface-bound IgG when TIGR4 and mutants were incubated with sera from 6AΔcps-colonized mice (Fig. (Fig.4D),4D), where PsaA was identified as the predominant cross-reactive antigen in Western blot assays. IgG induced by colonization of 23F bound to the surface of 6A but not to the surface of TIGR4. However, no loss of binding was seen with the absence of each cross-reactive antigen (data not shown), suggesting that none of these antigens are responsible for this surface-bound antibody. Therefore, PspA may induce a robust IgG response to the pneumococcal surface following colonization, whereas the responses to PpmA and PsaA are more limited. However, as suggested by Western blotting results, the cross-reactivity of each antigen varies depending on the colonizing strain.
Serum was taken from healthy adult volunteers and analyzed to determine if prior exposure to S. pneumoniae induced antibodies against the identified cross-reactive antigens. Since there were many reactive bands in Western blot assays, flow cytometry was used to detect surface antibody to the three candidate antigens. The amount of cross-reactive surface antibody against each antigen was calculated as the difference in binding between the wild type and mutants lacking the antigen. For PpmA, in 11/18 cases (0/6 sera tested against TIGR4, 5/6 sera tested against 6A, and 6/6 sera tested against 23F), ΔppmA mutants showed a significant decrease in human IgG binding (P < 0.05) (Fig. (Fig.5),5), based on comparisons of wild-type and mutant pairs in each serum sample. This result was not due to differences in capsule expression between wild-type and ppmA-deficient strains, as assessed by flow cytometry comparing binding of monoclonal anticapsular antibodies (data not shown). Surprisingly, the deletion of pspA did not result in a loss of surface IgG binding. As predicted based on the analysis of mouse sera, there was no loss of binding of human IgG in the absence of psaA (data not shown). These data suggest that most adult human sera contain antibodies recognizing surface-exposed epitopes of PpmA, with results varying depending on the target strain tested.
The in vivo roles of PspA and PpmA in cross-protection were investigated by colonizing mice with mutants lacking these antigens in the TIGR4 background and looking at mucosal cross-protection from 6A, since these two antigens were identified as cross-reactive using sera from TIGR4Δcps-colonized mice. Despite the robust immune response elicited against PspA and the broad cross-reactivity of PpmA, neither was necessary for cross-protection. Mice were still protected from 6A colonization (Fig. (Fig.6)6) and sepsis (data not shown) in the absence of PspA. Additionally, there was still protection from colonization after immunization with TIGR4 lacking both PspA and PpmA (Fig. (Fig.6).6). Since PsaA was identified as cross-reactive by sera from mice colonized with 6AΔcps, the role of PsaA was investigated by colonizing mice with the 6AΔpsaA mutant and looking at mucosal cross-protection from TIGR4. PsaA was not required for cross-protection, as mice lacking this antigen were still protected from TIGR4 colonization (Fig. (Fig.6).6). These data suggest that the three major cross-reactive antigens may contribute to, but are not necessary for, mucosal cross-protection.
The goal of this study was to identify pneumococcal surface antigens eliciting cross-reactive and cross-protective antibody responses during colonization. Experiments analyzing antibodies generated by murine colonization of three different pneumococcal isolates revealed several prominent cross-reactive antigens. However, no single surface antigen was identified as a broad target of cross-reactive antibody among all strains tested. Therefore, we focused on the most prominent cross-reactive antigens recognized in Western blot assays. Three out of four prominent bands recognized by cross-reactive antibody were found to be surface-associated proteins. Each of these three proteins, PspA, PpmA, and PsaA, has been previously proposed as a vaccine candidate or shown to induce protective responses when given as a purified protein together with an adjuvant (6, 12, 37), although evidence for cross-protection against colonization elicited by these purified antigens is more limited (6, 22). Our results provide an unbiased confirmation of the potential of each of these antigens to induce a cross-reactive immune response. However, following immunization with mutants lacking the relevant cross-reactive antigen(s), there was no significant decrease in cross-protection as determined by colonization density. This suggests that unknown minor antigens, which generate weaker cross-reactive responses, may mediate cross-protection and were not identified in our approach using Western analysis. Alternatively, there may be redundancy of cross-protective antigens for which our model cannot account. It is also possible that the serum IgG response is not representative of humoral protection at the mucosal surface and may be more important for preventing invasive infection. However, systemic immunization with PCV7 is known to induce IgG, which leaks into the nasal mucosa (35) and correlates with protection from carriage (10, 17). In this regard, we focused on IgG, since our data supported a role of antibody other than IgA in mucosal and systemic protection. Previous research has shown the importance of antibodies, as well as CD4+ T cells, in protection induced by colonization (39). The requirement for CD4+ T cells could be because of their contribution to humoral immunity. However, we cannot rule out the possibility that these, or other, antigens may be important for CD4+ T-cell-mediated, but not antibody-mediated, cross-protection, as demonstrated by Malley et al. (28). A further consideration is that levels of antibody induced by carriage may be much lower than those generated in response to PCV7 and, thus, may provide only limited cross-protection and require the combined effects of multiple antigens.
PspA was identified as one of the dominant targets of cross-reactive antibody induced by carriage. PspA, an abundant surface protein, is the most well studied and characterized pneumococcal protein vaccine candidate. When given systemically or mucosally as a purified antigen it can protect against both invasive disease and colonization in mouse models (7, 36, 42). PspA has a highly variable α-helical N terminus with sequence similarities in other regions that allow for classification into three families and six clades. Although there is thought to be a higher degree of protection elicited within the same family, protection has been seen between families (6, 31, 32). In our study, colonization by TIGR4Δcps elicited antibodies against PspA that cross-reacted with PspA from 6A. PspAs expressed by these two isolates have been sequenced and belong to different families, confirming that carriage may induce cross-reactivity between families. However, sera from 6AΔcps-colonized mice did not show prominent cross-reactivity with PspA of TIGR4, suggesting that cross-reactivity between families is variable and may also be influenced by differences in IgG induction between strains. Moreover, PspA was not found to be responsible for inducing high levels of cross-reactive antibody following colonization by another isolate of the same PspA family as 6A. Also, we were unable to detect cross-reactive antibody to PspA in normal human serum by using flow cytometry and comparing total IgG binding of parental and ΔpspA strains. Our approach to identify cross-reactive antigens by Western analysis was limited to antibodies detecting denatured proteins and would not identify conformational epitopes. However, the majority of sera induced by TIGR4Δcps colonization recognized PspA on the bacterial surface, suggesting that our approach accounted for the major targets of humoral immunity.
Another prominent target of cross-reactive IgG induced by colonization is PpmA. PpmA is a highly conserved lipoprotein studied for its potential as a vaccine candidate and is thought to be involved in the secretion and activation of cell surface molecules (37). Although the ability of PpmA to induce cross-protection has not been addressed in prior studies, antibodies generated against PpmA are cross-reactive against multiple strains of different serotypes, as seen in this study and others (37). Serum antibodies against PpmA are detected early in life and correlate with carriage. Antibodies to PpmA are lower in children with S. pneumoniae found in the middle ear (46) than in colonized controls, suggesting the immune response to PpmA may be protective. In our study, significant levels of anti-PpmA antibodies were detected in all adult normal human serum samples tested, although the level of surface-reactive anti-PpmA antibodies appeared to vary depending on the strain tested. This variation could be due to differential surface exposure of this antigen, or other interfering components, among strains. Whereas PspA has been confirmed to be surface exposed on encapsulated S. pneumoniae (11), it is not as clear whether PpmA and PsaA are masked by the capsule (18, 37). Our data suggest that the surface accessibility of PpmA may be strain specific.
A third prominent target of cross-reactive IgG induced by colonization is PsaA. PsaA is a highly conserved lipoprotein that is part of an ABC transporter involved in transport of manganese into the cell (13). PsaA is protective against carriage when given as a purified protein fused to cholera toxin (38). Western blotting identified PsaA as the major cross-reactive antigen in sera from 6AΔcps-colonized mice. As reported in other studies, we were unable to show surface reactivity of antibody to PsaA following murine colonization, or in normal human serum (18). PsaA is not required for mucosal cross-protection in our model, as mice are still protected in the absence of this antigen during immunization. However, it is worth noting that 6AΔpsaA colonizes inefficiently compared to the parental strain or other mutants in this background. As a result mice immunized with 6AΔpsaA induce less total anti-6A-specific antibody (data not shown). This may account for the partial protection following 6AΔpsaA compared to wild-type 6A immunization.
In summary, we analyzed the vaccine potential of colonization to understand how carriage may be an immunizing event, with the potential benefit of identifying proteins that elicit a protective immune response as candidates for a component vaccine. Previous reports have demonstrated that the identified cross-reactive antigens are sufficient, yet this murine colonization model shows they are not necessary, for mucosal cross-protection. Our data suggest that a protein-based vaccine that targets carriage may have to include multiple components to generate broad mucosal protection against this highly heterogeneous pathogen. Our findings also confirm the effect of live vaccination in protection from colonization (39), with significant cross-protection achieved even in the absence of the major candidate vaccine targets of humoral immunity. Broad protection is also an advantage of the killed whole-cell vaccination approach (26). These vaccination methods have the advantage of exposure of the immune system to multiple potential cross-protective antigens and may therefore overcome the limitations of narrow-range component vaccines.
We thank Kim Davis for expert technical assistance and also the proteomics core at the University of Pennsylvania. We also thank E. Ades (CDC, Atlanta, GA) for kindly providing the monoclonal antibody against PsaA.
This work was supported by grants from NIH (44231 and 38446) and The Bacterial Respiratory Pathogen Research Unit.
Editor: A. Camilli
Published ahead of print on 15 March 2010.