INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Streptococcus pneumoniae is a major infectious agent in humans and a significant cause of morbidity in the young, the elderly, and the immunocompromised (14, 16). Despite antibiotics, mortality due to pneumococcal bacteremia remains high (15). Of increasing concern is the growing number of antibiotic-resistant organisms among clinical isolates (3). Pneumovax, a polysaccharide vaccine for S. pneumoniae, does not elicit a B-cell memory response and is poorly immunogenic in those most at risk of serious disease (38, 42). Furthermore, there are at least 90 different serotypes of S. pneumoniae, and most of the protective anti-capsular polysaccharide antibodies are specific for a single serotype (20, 30). In contrast, the cell wall polysaccharide is relatively invariant across serotypes (43). Antibodies directed against phosphorylcholine (PC), a major epitope on the cell wall polysaccharide, protect against pneumococcal infections in mice (6, 23); the role of anti-PC antibodies in humans, however, remains controversial (13).

A growing number of reports describe the use of peptide mimics of carbohydrates and other nonprotein molecules in generating an immune response (2, 19, 21, 39, 40, 45, 48). In general, two approaches have been used to identify peptide mimics; peptide mimotopes either have been deduced from the interactions between anticarbohydrate antibodies and monoclonal anti-idiotypic antibodies or have been discovered by screening phage-displayed peptide libraries with monoclonal anticarbohydrate antibodies. Exploiting idiotype:anti-idiotype interactions to determine potential peptide mimics is attractive since the existence of an idiotypic interaction demonstrates that the antibody of interest is capable of interacting with both the cognate (nonprotein) antigen and proteins; however, only a subset of anti-idiotypic antibodies carry an internal image. Westerink et al. modeled an idiotypic interaction with an antimeningococcal antibody in order to design a peptide mimic of meningococcal carbohydrate (22, 48). Mice immunized with this peptide mimic were protected from bacterial infection, proving that an idiotypic interaction can be reduced to a peptide.

Immunization of BALB/c mice with PC induces one major idiotype, T15. T15 antibodies have the interesting and unique property of self-association. The self-association site has been shown to span a 24-amino-acid region in the PC binding site (ASRNKANDYTTEYSASVKGRFIVS; peptide 1). Kang et al. demonstrated that peptide 1 inhibited PC binding by binding in the antigen binding site; thus, peptide 1 was considered a potential peptide mimic of PC (24). We tested a nested set of 12-mer peptides (peptides 2 to 8, spanning amino acids 1 to 12, 3 to 14, 5 to 16, 7 to 18, 9 to 20, 11 to 23, and 13 to 24, respectively) for their ability to block binding of PC by a panel of anti-PC monoclonal antibodies in order to determine the minimal peptide size and the specific 24-mer peptide residues needed for mimicry. C-terminal peptides (i.e., peptides 6, 7, and 8, beginning at amino acids 9, 11, and 13, respectively) were shown to bind at or near the PC binding site.

Based on these data, we immunized mice with peptide 1, 7, or 8 coupled to bovine serum albumin (BSA). Peptide 7- and peptide 8-BSA failed to elicit anti-PC antibodies. In contrast, the anti-PC antibodies elicited by peptide 1-BSA expressed the canonical T15 heavy chain but did not appear to express the canonical T15 light chain, thus demonstrating that peptide 1-BSA elicited an idiotypic profile different from that elicited by PC-BSA. PC-BSA and peptide 1-BSA elicited antibodies of different isotypes. Peptide 1-BSA elicited a significant anti-PC immunoglobulin G (IgG) response consisting of mainly IgG1 and some IgG_2a; no IgM nor IgG3 activity was noted. In contrast, PC immunization elicits mainly IgM, IgG3, and some IgG1 (8, 11). Interestingly, mice immunized with peptide 1-BSA were not protected from lethal pneumococcal infection even though anti-PC antibodies expressing the T15 heavy chain were present. The inability of peptide 1-BSA to elicit the canonical T15 light chain may account for this lack of protection. This idiotype-derived mimotope of PC will serve as a lead compound for the development of protective, T-cell-dependent vaccines for S. pneumoniae and other PC-containing pathogens and will be a useful tool for gaining an understanding of both immunologic cross-reactivity and the structural requirements for immune protection.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Peptides with N-terminal acetates and C-terminal amides were synthesized by Research Genetics (Huntsville, Ala.). BSA, glutaraldehyde, and PC chloride were purchased from Sigma (St. Louis, Mo.). PC-BSA was synthesized according to the method of Chesebro and Metzger (7). Mice were purchased from Jackson Laboratory (Bar Harbor, Maine). Secondary antibodies were purchased from Sigma, Southern Biotech (Birmingham, Ala.), or Zymed (South San Francisco, Calif.). Rat anti-T15 monoclonal antibodies T139 and TC54 were generous gifts from Matthew Scharff.

Conjugation. BSA (5 mg) was dissolved in a 0.1 M sodium citrate solution (pH 5.5; 500 µl) and mixed with peptide (1, 7, or 8; 5 mg) in 0.1 M sodium citrate (pH 5.5; 500 µl) to provide a BSA:peptide ratio of 1:25 (for peptide 1) or 1:50 (for peptides 7 and 8). Glutaraldehyde (0.1%) was added, and the solution was incubated for 1 h at room temperature. The reaction mixture was dialyzed against phosphate-buffered saline (PBS) for 5 days at 4°C.

Immunizations. Members of groups of 6-week-old female BALB/c or CBA/N mice (Jackson Laboratories) were initially immunized with 100 µg of the peptide- or PC-BSA conjugate, or with BSA alone, in complete Freund's adjuvant H37Ra (DIFCO); for the booster immunizations, performed on day and day 42, incomplete Freund's adjuvant was used. The mice were bled before each immunization, 2 weeks after the final immunization, and 1 week before pneumococcal infection.

Antibody purification. The day 57 postimmunization sera from peptide-BSA-immunized mice were pooled, diluted with an equal volume of phosphate buffer (0.1 M, pH 8), and batch adsorbed with PC-Sepharose (Pharmacia, Piscataway, N.J.). Bound antibodies were eluted with PC chloride (10 mM in Tris-buffered saline) and dialyzed against PBS overnight at 4°C to remove bound PC.

ELISAs. For enzyme-linked immunosorbent assays (ELISAs), microwells were coated with antigen overnight at 4°C, using a 20-µg/ml solution of PC-BSA or BSA or a 5-µg/ml solution of C polysaccharide (Statenserum Institut, Copenhagen, Denmark). The T15-positive monoclonal antibodies PC2 (µ, 2a, and 2b), PC1.4.1 (1), and M4.37 (3) were used to generate standard curves. Isotype-specific or total IgG goat anti-mouse secondary antibodies were used for ELISA development. Peptides were coated at a concentration of 10 µM, and peptide DRIPMDYWGQGTSVTVSS was used as a control.

Opsonization. Serum samples were threefold serially diluted in opsonization buffer (1.2 mM CaCl₂, 0.5 mM MgCl₂, 0.1% gelatin, and 10% fetal bovine serum in Hanks' balanced salt solution) in round-bottom microwells (Costar, Cambridge, Mass.). Ten microliters (about 100 CFU) of a pneumococcus strain SP85 (serotype 6A) culture was added to each of the wells, and the plates were incubated for 15 min at 37°C. Ten microliters of rabbit complement (Pelfreeze) and 40 µL of a HL-60 cell suspension (differentiated to granulocytes with dimethyl formamide; 10⁷ cells/ml in opsonization buffer [9]) were added to each well, and the plates were incubated at 37°C for 45 min. The plates were subsequently incubated at 37°C in a candle jar for 7 h, and the resulting colonies were counted. The opsonization titer was the reciprocal of the final serum dilution in the well which gave a 50% reduction of colonies.

Protection assay. WU-2, a virulent type 3 strain of S. pneumoniae (a gift from D. Briles), was used in the protection assays. Bacteria were streaked out on a blood agar plate (Becton Dickinson, Franklin Lakes, N.J.) and incubated for 18 h at 37°C in a 5% CO₂ atmosphere. An inoculum broth culture was prepared by incubating 5 to 10 colonies in Todd-Hewitt broth until the OD at 620 nm was 0.25. One-milliliter aliquots were frozen and stored at 70°C until needed. The inoculum broth was thawed and diluted in Todd-Hewitt broth for challenge. BALB/c and CBA/N mice, either naïve or immunized, were challenged intraperitoneally (i.p.) with 100 or 10 CFU, respectively. The number of CFU in the inoculum was confirmed by plating on blood agar plates.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

C-terminal portions of a T15-derived peptide inhibit binding of PC to a panel of anti-PC monoclonal antibodies. Antibodies expressing the T15 idiotype dominate the response to PC (28). These antibodies have been shown to self-associate (25). Kang et al. demonstrated that a 24-mer peptide (ASRNKANDYTTEYSASVKGRFIVS; peptide 1) spanning a portion of CDR2 and FR3 of the H chain bound at or near the PC binding site, suggesting that it might be a potential mimic of PC (24). To test this hypothesis and to determine the minimum size and specific amino acid residues needed for mimicry, we tested a nested set of 12-mer peptides (peptides 2 to 8, spanning amino acids 1 to 12, 3 to 14, 5 to 16, 7 to 18, 9 to 20, 11 to 23, and 13 to 24, respectively) for their ability to inhibit binding of PC by a panel of anti-PC monoclonal antibodies. Three of the anti-PC monoclonal antibodies, PC.1.4.1 (1), PC2 µ (µ), and M4.37.3 (3), have the canonical T15 sequence, with M4.37.3 differing from the germ line sequence at four amino acid residues (29, 44). A fourth monoclonal anti-PC antibody, 180.2E3.4 (1; a gift from J. L. Claflin) is of the M603 type and expresses the T15 heavy chain but not the light chain (1).

Immunization of mice with T15-derived peptide mimics of PC. Based on the inhibition data, we immunized BALB/c mice with a BSA conjugate of peptide 1, 7, or 8, PC-BSA, or BSA. The use of different secondary antibodies allowed a quantitative determination of relative amounts of anti-PC antibody within an assay but not a comparison of absolute amounts between different assays. Neither peptide 7-BSA nor peptide 8-BSA was immunogenic (data not shown). Peptide 1-BSA induced an IgG PC-binding response significantly higher than that caused by BSA alone (Fig. 1). The antibodies induced by peptide 1-BSA also bound cell wall C polysaccharide, thus demonstrating their antibacterial activity (Fig. 2) (23, 43).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Peptide 1-BSA induces an anti-PC IgG response. Preimmunization sera from groups of mice were pooled, and postimmunization sera from individual mice were titrated on microtiter wells coated with PC-BSA. The amount of anti-PC antibody was determined from standard curves generated by using the T15-expressing anti-PC IgG1 monoclonal antibody PC1.4.1 and a goat anti-mouse total IgG secondary antibody PC1.4.1 and a goat anti-mouse total IgG secondary antibody. Symbols represent postimmunization sera of individual mice. A Mann-Whitney U test returned a P value of 0.0396 for peptide 1-BSA-immunized mice compared to BSA-immunized mice. d, day.

View larger version (10K): [in this window] [in a new window]   FIG. 2.   Peptide mimics of PC induce a C polysaccharide binding response. See Materials and Methods for an explanation of the binding units. Symbols represent individual mice. d, day.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   PC affinity purification of postimmunization anti-peptide 1-BSA serum isolates antipeptide activity. Binding to the cognate peptide or a control peptide (DRIPMDYWGQGTSVTVSS) was tested at a 1:500 dilution of whole serum. Affinity-purified antibodies were used at 550 ng/ml; non-PC-binding antibodies were tested at 550-ng/ml and 2.5 µg/ml. Error bars represent standard deviations. , peptide control (1 µM); , peptide 1 (1 µM). d, day.

View larger version (11K): [in this window] [in a new window]   FIG. 4.   Peptide 1-BSA induces the T15 canonical heavy chain. Peptide 1-BSA induces more TC54 reactivity than BSA. The preimmunization sera from groups of mice were pooled together. Symbols represent postimmunization sera of individual mice.

View larger version (11K): [in this window] [in a new window]   FIG. 5.   The anti-PC response induced by peptide 1-BSA consists of IgG1. Preimmunization sera from groups of mice were pooled, and postimmunization sera from individual mice were titrated on microtiter wells coated with PC-BSA. The amount of anti-PC antibody was determined from standard curves generated by using the T15-expressing anti-PC IgG1 monoclonal antibody PC1.4.1 and a goat anti-mouse IgG1 secondary antibody. The symbols represent postimmunization sera of individual mice.

Immunization of CBA/N mice with peptide 1-BSA. CBA/N mice carry the X-linked immunodeficiency defect and do not respond to T-cell-independent type II antigens (41). As a result, they do not mount a T15 response to PC immunization (33). CBA/N mice were immunized with peptide 1-BSA in order to determine if peptide 1 was capable of overcoming the immune defect in these mice and eliciting a PC binding response similar to that seen in BALB/c mice. Peptide 1-BSA induced IgG anti-PC activity but no expression of the T15 heavy chain (data not shown).

PC-binding antibodies elicited by peptide-BSA conjugates neither opsonized bacteria nor were protective. None of the sera taken from BALB/c mice on day 63 following three immunizations with any of the peptide- or PC-BSA conjugates was able to opsonize S. pneumoniae (data not shown). Serum from mice immunized with PC-BSA does not always opsonize bacteria in vitro even though the mice are protected from lethal pneumococcal infection. The bacterial strain used for the opsonization experiments is an encapsulated strain that does not contain PC in its capsular polysaccharide. Therefore, it is likely that the PC epitopes on C polysaccharide were unavailable for antibody binding. Neither BALB/c nor CBA/N mice immunized with peptide 1-BSA were protected from lethal pneumococcal infection (data not shown). Thus, peptide 1-BSA elicits PC-binding antibodies that are not protective. The inability of peptide 1-BSA to elicit the canonical heavy- and light-chain combination may explain why this PC-binding response fails to protect mice from bacterial challenge. It has been demonstrated that T15-positive antibodies of all isotypes are protective against pneumococcal infections (4, 5), while alterations of the canonical T15 idiotype have been shown to decrease affinity for PC as well as protective potential (26, 35).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have demonstrated that peptide 1, a 24-mer peptide derived from the heavy-chain CDR2 and FR3 (24), as well as C-terminal peptides 6, 7, and 8 (i.e., amino acids 9 to 20, 11 to 22, and 13 to 24) were able to inhibit the binding of anti-PC monoclonal antibodies to PC-BSA. Thus, all were potential mimics of PC. Earlier, Lai et al. reported the inhibition of PC binding to antibodies using peptides derived from the heavy-chain CDR1 (27). These authors concluded that the peptides they tested inhibited the binding of PC in a nonspecific manner since several control peptides of similar size also inhibited PC binding. The peptides used in our work have related sequences and are of similar size yet exhibited different abilities to inhibit PC binding. Thus, these CDR2-derived peptides interact with the anti-PC antibodies specifically. We do not know why the pattern of peptide inhibition differed for each of the antibodies. It may be that peptide 1 has more secondary structures such that it binds less well in the antigen binding site. It is also tempting to speculate that the difference is due to isotype because PC2 and T15 have identical variable-region sequences (29, 44), a fact demonstrated by using anti-idiotypic reagents. Anti-idiotypes have been reported to display differential reactivity with T15 antigens depending on the heavy-chain isotype (34). If isotypes display subtle differences in antigen binding, it may help to explain the isotype restriction that is seen in the anti-PC response (11).

Westerink et al. were able to model a peptide mimotope of Neisseria meningitidis group C based on an idiotypic interaction (48). Several groups have used phage-displayed peptide libraries to isolate peptide mimics of bacterial carbohydrates, with various degrees of success. Pincus et al. and Grothaus et al. isolated peptides that elicit antibacterial responses directed against group B streptococci and N. meningitidis group A, respectively (17, 40). Phalipon et al. isolated 19 peptide mimics of Shigella flexneri lipopolysaccharide, of which only four elicited lipopolysaccharide-binding antibodies (39). De Bolle et al. and Moe et al. isolated peptide mimics of Brucella spp. and N. meningitidis group B, respectively, which elicited only very weak antibacterial responses (10, 32). Valadon et al. isolated peptide mimics of Cryptococcus neoformans which elicited the correct idiotype but did not bind C. neoformans (46). Thus, binding alone does not guarantee that a peptide mimic will successfully elicit the desired immune response.

Peptides 7 and 8 bound at or near the PC binding site with affinities similar to that of peptide 1, but they did not induce a similar PC-binding response. It may be that peptides 7 and 8 occupy only part of the PC binding site, enough to inhibit PC binding but not enough to be a mimotope.

Peptide 1-BSA elicits the canonical heavy chain but not the canonical light chain, suggesting an interaction with the heavy chain alone. Valadon et al. have reported analogous results for a peptide mimic of C. neoformans (46). A crystal structure of an anticryptococcal monoclonal antibody bound to the peptide mimic revealed that the peptide occupied only part of the binding site, making contacts mainly with the heavy chain (49). Similarly, anti-idiotypic monoclonal antibodies raised against an anti-Brucella carbohydrate monoclonal antibody fail to elicit a carbohydrate binding response (50). Young et al. suggest that the anti-idiotypic monoclonal antibodies do not mimic the carbohydrate because they do not fill the carbohydrate binding site of the original antibody (50). A crystal structure of a monoclonal anti-PC antibody complexed with one of the peptide mimics would provide information about the molecular interactions between the antibody and the peptide mimics. Moreover, an anti-PC antibody:peptide mimic structure would provide information on how to modify the peptide in order to elicit a protective anti-PC response.

Earlier work by McNamara et al. used 4C11, a monoclonal anti-idiotypic antibody raised against a monoclonal anti-PC antibody, coupled to keyhole limpet hemocyanin to induce a protective PC-binding response in mice (31). Approximately 90% of the PC-binding antibodies were T15 positive. Peptide 1 was derived from a different idiotypic interaction and has no sequence similarity to 4C11. Immunization with peptide 1-BSA elicited an anti-PC response without induction of canonical T15 antibodies. Only the T15 heavy chain occurred in the response. These antibodies failed to protect mice from a lethal S. pneumoniae infection. Bacterial challenge of BALB/c mice occurred on day 170 after the total amount of anti-PC antibody, as well as T15-positive anti-PC, had peaked, as seen in Fig. 1; however, the amount of anti-PC antibody present was still within the range of reported protective anti-PC titers (6, 47). Moreover, sera taken on day 63 (when the anti-PC titer was high) from mice immunized with each of the peptide-BSA conjugates failed to opsonize bacteria in vitro. Thus, the lack of protection afforded by peptide 1-BSA is likely due to a failure to elicit the protective idiotype rather than to a low anti-PC titer (18). This is consistent with previous data suggesting that the heavy-chain CDR3 sequences affect, or the associated light chain determines, bacterial specificity (18).

It is interesting that the anti-PC antibody does bind C polysaccharide despite its failure to protect mice against bacterial infection. It has previously been reported that both protective and enhancing antibodies may occur in an immune response and that these antibodies differ with respect to the antigenic epitopes to which they bind (36). It has also been reported that monoclonal antibodies directed against the same epitope of cryptococcal polysaccharide may differ in their ability to protect mice against a lethal cryptococcal infection (37, 51-53). Understanding the nature of a protective epitope and a protective antibody is complex and a major challenge in vaccine development.

PC is considered to be a T-cell-independent type II antigen and elicits mainly IgM, IgG3, and some IgG1, even though all isotypes of anti-PC antibodies have been shown to be equally protective (5). The IgG anti-PC response elicited by peptide 1-BSA consisted of mainly IgG1; no IgG3 or IgM activity was detected. The presence of IgG1 anti-PC antibodies suggests that peptide 1-BSA elicits T-cell help, perhaps through T-cell recognition of variable-region epitopes. Thus, the peptide mimics may induce antipneumococcal responses in children and other populations whose members do not respond well to T-cell-independent vaccines such as Pneumovax.

Our work proves that peptides can mimic PC, a small nonpeptide epitope found on several major pathogens. It will be important to gain further understanding of the role of both idiotype and the isotype in a protective anti-PC response. It is clear that one can develop peptide mimics of PC, but the idiotypic and isotypic characteristics and protective potential elicited will differ.