INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Neisseria meningitidis is the most common cause of bacterial meningitis in the United States, and a meningococcal vaccine is currently licensed in the United States. However, this vaccine needs to be improved for various reasons. First, this vaccine contains capsular polysaccharides (PS) from N. meningitidis serogroups A, C, Y, and W135 but not serogroup B (11). The lack of protection against N. meningitidis group B (NMGB) in a meningococcal vaccine is a serious shortcoming, because NMGB may account for 50% or more of all meningococcal meningitis cases in Europe and North America (3, 29). Second, this vaccine does not elicit antibodies in young children, who account for about 50% of meningococcal meningitis cases (29). Although conjugation of the meningococcal PS to protein carriers makes group A, C, and Y capsular PS immunogenic in young children, group B PS-protein conjugate remains poorly immunogenic (19).

There are several obstacles to generating a vaccine effective against NMGB. One obstacle is that NMGB PS may elicit autoantibodies. Antibodies to NMGB PS can be generated after natural infection (25) or after immunization with a chemically modified NMGB PS (18) or Escherichia coli K92 PS (9), which is a polymer of sialic acid (PSA) with alternating (2-8) and (2-9) linkages. However, the antibodies were found to bind frequently to both NMGB PS and neuronal tissue (15, 25). This cross-reaction occurs because both express a linear (2-8) PSA. NMGB PS is a PSA with about 200 repeating units (12), and neuronal tissue has the same but shorter (about 10 to 50 repeating units) PSA as a part of neuronal cell adhesion molecule (NCAM) (8). Although the notion is controversial, the antibodies cross-reacting with the neuronal PSA are thought to have the potential to cause neurological damage. Another major obstacle is the absence of simple alternative NMGB vaccine candidate antigens. For instance, outer membrane proteins have been used as a vaccine, but this approach is limited because of significant serologic heterogeneity among different strains of NMGB (3).

Two new approaches for generating an NMGB vaccine have been suggested. One approach is to find a new vaccine candidate molecule. This approach received a significant boost from the sequencing of the entire genome of NMGB (26). Another approach is to use peptides that mimic the bacterium-specific epitope of NMGB PS as the vaccine. The feasibility of this approach has been demonstrated with the evaluation of peptide mimics of meningococcus group C (33). This approach has now become more amenable with the development of phage display technology (10, 30), which can be used to identify peptide mimotopes of PS (31). We now report the development of monoclonal antibodies (MAbs) that bind and kill NMGB without binding neuronal PSA and use of these MAbs to identify peptide mimotopes of NMGB capsular PS.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Antigens and bacteria. Various strains of bacteria used for this study are summarized in Table 1. All E. coli strains were cultured in Luria-Bertani (LB) broth or on LB agar plates. Neisseria strains were grown on chocolate agar plates in a candle jar. To obtain a large number of Neisseria bacteria with minimal biohazard, many chocolate agar plates were plated with the bacteria and the bacteria were then harvested from the plates after a 6-h incubation at 37°C. All bacteria were aliquoted in Hanks' balanced salt solution (HBSS) containing 20% glycerol and stored at 70^oC.

Enzyme-linked immunosorbent assay (ELISA). Ninety-six-well microtiter plates (Nunc, Roskilde, Denmark) were coated with various antibody capture agents as described below. The plates were then washed with 0.05% Tween 20 in phosphate-buffered saline (PBST) and blocked with 2% skim milk in PBST. Appropriately diluted samples were added to the plates and incubated for 1.5 h at 37^oC. After washing, alkaline phosphatase-labeled goat anti-mouse immunoglobulin (Ig; Sigma) was added. After a 90-min incubation, p-nitrophenyl phosphate was added to the plates and optical densities of the plates were measured at 405 nm.

Production of MAbs. BALB/cByJ mice from Jackson Laboratory (Bar Harbor, Maine) were intraperitoneally immunized with E. coli K1 four times, twice with 10⁷ CFU, on days 0 and 4, and twice with 10⁸ CFU, on days 7 and 10. Live bacteria were used for fusions 1, 2, and 3, but heat-killed bacteria were used for fusions 4 and 5. Spleens were harvested on day 13 for fusion with Sp2/0-Ag14 myeloma cells, as described previously (24). The resulting hybridoma supernatants were screened by a sandwich-type ELISA as described above using ELISA plates coated with purified E. coli K1 PS in the first test and plates coated with killed NMGB bacteria in the second test. Only the hybridomas producing antibodies binding to both NMGB bacteria and E. coli K1 PS were cloned. Twelve hybridoma clones, all immunoglobulin M  [IgM()], were produced from five fusions.

Flow cytometry. Flow cytometry was used to test MAbs for binding to various strains of the bacteria, listed in Table 3. The bacteria were killed by incubation in 2% formaldehyde for 1 h, washed with HBSS, and suspended at 5 × 10⁸ CFU/ml in HBSS containing 1% glycine in order to neutralize any remaining formalin. Twenty microliters of killed bacteria (5 × 10⁸ CFU/ml) and 50 µl of MAb (5 µg/ml) were mixed in a well of a microtiter plate. An irrelevant IgM MAb was used as a negative control. After 1 h of incubation, this mixture of bacteria and MAbs was washed and suspended in 100 µl of fluorescein isothiocyanate-conjugated anti-mouse Ig (Sigma). After 1 h of incubation, the bacteria were washed again and fixed by resuspension in 400 µl of PBS containing 0.5% paraformaldehyde. Fluorescence, forward scatter, and side scatter of the bacteria were measured with a flow cytometer (FACScaliber; Becton Dickinson, Mountain View, Calif.).

Bactericidal assay. The bactericidal assay was performed in 96-well microtiter plates (Corning Inc., Corning, N.Y.) with modifications of a previously described procedure (27). Briefly, 30 µl of bacterial suspension containing 2,500 CFU, 50 µl of appropriately diluted antibiotic-free antibody, and 20 µl of baby rabbit complement or agammaglobulinemic human serum were added to each well. The concentration of complement was 2 to 20% depending on the susceptibility of the bacteria to the complement, and an isotype-matched irrelevant MAb was used as a negative control. Target bacterial strains are listed in Table 1. Two different strains of NMGB (ATCC 13090 and H44/76) were used to avoid killing by binding to antigens that were not capsular PS. After a 1-h incubation with shaking, 10 µl of the reaction mixture was plated on a chocolate agar plate. CFU were determined after an overnight incubation at 37°C in a candle jar. Each assay was performed in triplicate, and the mean of the triplicate was used to calculate the percentage of killing by the formula [(CFU_{no antibody}  CFU_sample)/ CFU_{no antibody}] × 100. This assay was repeated at least two times.

Histochemistry. The staining was performed with the brains obtained from 1- to 2-week-old C57BL/6 mice. Five-micrometer-thick cryosections of the brain were incubated with 1 to 2 µg of MAb/ml for 75 min at RT. The slide was washed with PBS and incubated with biotin-conjugated goat anti-mouse IgM antibody (Vector Laboratories, Burlingame, Calif.) for 30 min at RT. The slide was washed again with PBS and incubated with streptavidin-peroxidase (Jackson ImmunoResearch Laboratory, West Grove, Pa.) for 30 min. The slide was developed with a peroxidase substrate, 3-amino-9-ethylcarbazole, from Scy Tek (Logan, Utah) and counterstained with Mayer's hematoxylin stain (Poly Scientific, Bayshore, N.Y.).

Production of phage clones expressing the mimotopes. Two phage libraries from New England Biolabs Inc. (Beverly, Mass.) were used for our study. One contained a linear peptide with 7 amino acids, and the other contained a circular peptide with 7 amino acids between two cysteines. The two cysteines were joined by a disulfide bond to form a circular peptide. The phage library was biopanned with purified MAbs HmenB1, HmenB2, HmenB3, HmenB13, and HmenB14, as described by the manufacturer of the library. Briefly, MAb-coated 60-mm-diameter culture dishes (Nunc) were prepared by incubating the dish with 1.5 ml of 0.1 M NaHCO₃ (pH 8.6) containing 100 µg of MAb/ml overnight at 4°C, washing with 0.1% Tween 20-Tris-buffered saline (0.1% TBST; 50 mM Tris-Cl [pH 7.5], 150 mM NaCl), and blocking with 0.5% bovine serum albumin in 0.1 M NaHCO₃. Phage binding to the MAb was enriched by incubating the phage library in the dish, gently washing away unbound phage, and removing the phage bound to the dish by elution with 0.2 M glycine-HCl (pH 2.2) and rapidly neutralizing the phage with 1 M Tris-Cl (pH 9.1). Bound phage was expanded and subjected to additional cycles of enrichment as described above. After three cycles of enrichment, bound phage was cloned. Individual clones of phage were expanded for further studies.

Binding and competitive binding assays of phage on antibody-coated plates. Ninety-six-well microtiter plates (Nunc) were coated with MAbs by adding 100 µl of a carbonate buffer (pH 9.6) containing 10 µg of MAb/ml. After an overnight incubation, the plates were washed and blocked with 2% skim milk in PBST. Fourfold serial dilution of the solutions containing amplified phage (1 × 10¹¹ to 10 × 10¹¹ PFU/ml) was performed, and the serially diluted phage was added to the plates. In some cases, to test for the specificity of phage binding to MAb, a serial dilution of E. coli K1 PS (from 1,600 to 0.2 µg/ml) or heat-inactivated NMGB (1.6 × 10¹⁰ to 2 × 10⁵ cells/ml) was added to the well along with a fixed number of phage (10¹⁰ PFU/ml). After a 1.5-h incubation at RT with gentle agitation, the plates were washed and loaded with peroxidase-labeled antiphage MAb (Pharmacia). One and one-half hours later, the plates were washed and loaded with tetramethylbenzidine (Kirkegaard & Perry Laboratories, Gaithersburg, Md.). The color development was stopped, and the plates were read at 450 nm.

DNA sequencing of phage peptide motif. To determine the DNA sequence of the peptide mimotope, phage clones were expanded by growing them in 1-ml E. coli cultures for 4 to 5 h at 37°C. To recover the phage, the bacteria were removed from the culture suspension by a brief centrifugation (10,000 × g for 10 s), 500 µl of the culture supernatant was mixed with 200 µl of 20% polyethylene glycol-2.5 M NaCl solution, and the mixture was centrifuged (10,000 × g) for 10 min. The pellet containing the phage was resuspended with 100 µl of a Tris-EDTA (TE) buffer with iodide (10 mM Tris-Cl [pH 8.0], 1 mM EDTA, 4 M NaI) and 250 µl of absolute ethanol. The phage was then washed with 70% ethanol, dried, and resuspended in 30 µl of TE buffer. Five microliters of the phage suspension was subjected to dideoxy termination reactions using a DNA sequencing kit (Perkin-Elmer, Norwalk, Conn.), 96 sequencing primer from New England Biolabs, and AmpliTaq DNA polymerase FS. The sequence of the mimotope was obtained by running the above reaction products through an automated DNA sequencer from Applied Biosystems (Foster City, Calif.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Characterization of antigens for the MAb. From five mice immunized with E. coli K1 bacteria (strain RS218), we obtained 12 hybridomas expressing IgM heavy and  light chain isotypes. Consistent with their selection prior to cloning, these hybridoma antibodies bound to the ELISA plates coated with NMGB bacteria and purified E. coli K1 PS (data not shown). The preparation of E. coli K1 PS was highly purified and was not likely contaminated with other antigens (W. Vann, personal communication). In addition, the 12 hybridoma antibodies bound to EV36 (Table 2), which is an E. coli K-12 strain possessing the genes for K1 capsular PS production (32) and expresses nonacetylated (2-8) PSA just like NMGB (R. Silver, personal communication). These findings taken together strongly suggest that all the MAbs bind to (2-8) PSA and/or to a structure invariably associated with the PS (e.g., the lipid anchor of the capsular PS [4, 14]).

Autoreactivities of the MAbs. Since antibodies to NMGB PS often cross-react with human (2-8) PSA expressed in neural tissue, cross-reactivity of our MAbs with human PSA was initially assessed with a human neuroblastoma cell line, CHP-134 (21). CHP-134 expresses NCAM decorated with a 50-mer of PSA (21) and can serve as a source of human PSA antigen. More than 90% of the CHP-134 cells used for our study expressed a marker of NCAM, CD56 (Fig. 1A), indicating that our CHP-134 cells expressed NCAM. Indeed, HmenB1, HmenB4, HmenB5, HmenB8, HmenB9, and HmenB10 had weak or moderate binding to CHP-134 cells at approximately 10 mg/liter (Table 2 and Fig. 1B). However, HmenB2, HmenB3, HmenB6, HmenB7, HmenB13, and HmenB14 did not bind to CHP-134 cells detectably under the conditions used here (Table 2 and Fig. 1B). For instance, HmenB2, HmenB3, HmenB13, and HmenB14 did not bind even at a high concentration (25 mg/liter). Thus, some of the MAbs are specific for the bacterial PSA and do not cross-react with human PSA.

View larger version (61K): [in this window] [in a new window]   FIG. 1.   (A and B) The number of CHP-134 cells (y axis) versus cellular fluorescence (x axis). (A) CHP-134 cells were stained with anti-CD56 (green line) or no antibody (dotted line). (B) Cells were stained with 25 µg of HmenB1 (red line) or HmenB3 (green line) per ml or no antibody (dotted line). (C and D) A cryosection of 10-day-old mouse brain (cerebellum) was stained with HmenB3 (C) and HmenB1 (D). Positive-staining cells have a brick-red color. Approximate magnifications, ×57 (C) and ×110 (D).

Characterization of bactericidal activity of the MAb. Antibodies may bind to bacteria without promoting the bactericidal activity of complement (15). Thus, the MAbs were tested at the concentrations of 5 and 0.5 mg/liter for killing N. meningitidis group A (M239), B (ATCC 13090 and H44/76), C (BB-305), and Y (S-1975) bacteria in the presence of rabbit or human complement. Similar results were obtained with both antibody concentrations, and results obtained at 0.5 mg/liter are shown in Table 3. No MAbs killed N. meningitidis groups A, C, and Y to any detectable degree even at a high (5-mg/liter) concentration, except for HmenB5. HmenB5, which bound group C meningococci very well, readily killed group C meningococci in the presence of rabbit complement (Table 3).

View larger version (20K): [in this window] [in a new window]   FIG. 2.   Bactericidal activities (y axis) at different concentrations of MAb (x axis). Individual curves show the bactericidal activities against NMGB (ATCC 13090) by HmenB3 (squares), HmenB13 (circles), and HmenB14 (triangles). All the curves showing the bactericidal activities against N. meningitidis groups A (strain M239), C (strain BB-305), and Y (strain S-1975) by the three MAbs are inside the shaded triangle.

Selection of phage clones and analysis of peptide sequence motif. The above data taken together strongly suggested that six MAbs bind to (2-8) PSA (and/or its associated structures) from NMGB and can promote killing of the bacterium by complement but do not bind the (2-8) PSA from neural tissue. We therefore chose four (HmenB2, HmenB3, HmenB13, and HmenB14) of the six hybridomas as well as HmenB1 (a negative control) to isolate peptide mimotopes. After three rounds of biopanning a phage library displaying random linear and circular peptides, many phage clones binding to ELISA wells coated with the five MAbs were identified. A randomly chosen irrelevant phage (Pha-CNTL) clone bound poorly to these wells (Fig. 3).

View larger version (25K): [in this window] [in a new window]   FIG. 3.   The amount of bacteriophage bound (y axis) to the microwells coated with different MAbs versus various dilutions of phage clones (x axis). Dilution (x axis) indicates fold dilutions, and the optical density (y axis) is expressed in milli-absorbance units. PhaB3L1 (filled circles), PhaB3C1 (open circles), and Pha-CNTL (inverted open triangles) were tested with HmenB3-coated microwells. PhaB1L1 (filled squares) was tested with HmenB1-coated microwells.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   (A) The amount of bacteriophage bound (y axis) to the microwells coated with different MAbs versus various concentrations of E. coli K1 PS in solution (x axis). The optical density (y axis) was expressed in milli-absorbance units. The concentration of phage used for the experiment was 1010 PFU/ml. PhaB3L1 (filled circles) and PhaB3C1 (open circles) were tested with HmenB3-coated microwells. PhaB1L1 (filled squares) was tested with HmenB1-coated microwells. (B) Cross-reactive binding of the phage clones to microwells (y axis) versus the dilution of the phage stocks (x axis). Dilution (x axis) indicates fold dilutions, and the optical density (y axis) is expressed in milli-absorbance units. Individual curves indicate the binding of PhaB3L1 to HmenB3-coated microwells (filled circles), PhaB3L1 to HmenB1-coated microwells (open circles), PhaB1L1 to HmenB1-coated microwells (filled squares), and PhaB1L1 to HmenB3-coated microwells (open squares).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Due to the clinical importance of NMGB, the binding specificity of antibodies to its capsular PS has been studied extensively. The antibodies have been found to differ in their abilities to kill NMGB bacteria in the presence of complement, in their binding to NMGB PS of various lengths, and in their cross-reactivities to various antigens including neuronal cells. For instance, Jennings and his colleagues found that N-propionyl derivatization of NMGB reduces the induction of antibodies binding the host NCAM and described the binding specificity diversity among the hybridomas specific for N-propionylated NMGB (27). A recent study by Granoff et al. (15) showed that hybridomas obtained from mice immunized with N-propionylated NMGB differ in complement-dependent killing of NMGB bacteria and binding to NMGB PS. Furthermore, they found that some MAbs are autoreactive since they bind to NCAM expressed on the CHP-134 cell line.

Although our MAbs were obtained by immunizing mice with E. coli K1, they largely reflected the antigen binding pattern described above. We have also obtained several autoreactive hybridomas. For instance, HmenB1 readily bound the CHP-134 cells and the brain tissue. Even though it readily bound to NMGB PS, it was not bactericidal against NMGB. Another autoreactive antibody was found to display unusual binding properties: in addition to binding NMGB, HmenB5 bound to the PS of group C N. meningitidis and killed this bacterium. This type of antibody specificity may be produced with E. coli K92 immunization, but one MAb reported in the literature with the identical binding property was produced with NMGB immunization (20). Group C PS in the meningococcal vaccine used in the U.S. military may elicit this type of cross-reactive antibody, and such an antibody may be responsible for the observed lack of increase in group B serotype outbreaks in the U.S. military (5). The epitope bound by HmenB5 must include the PSA because this MAb binds to the CHP-134 cell line and to purified E. coli K1 PS. Interestingly, the mice carrying the autoreactive hybridomas in their peritoneum did not show any obvious neurological symptoms, consistent with previous observations of patients with a high level of anti-NCAM antibodies due to myelomas or meningococcal infections (25).

In addition to the autoreactive antibodies, however, we also obtained many MAbs recognizing a bacterium-specific epitope(s). The bacterium-specific hybridomas bound to the E. coli strain EV36. This finding indicates that an unrelated (K-12) strain of E. coli can gain the bacterium-specific epitope with the transfer of the gene only for the capsular PS of E. coli K1. Also, the bacterium-specific epitope was present on the purified E. coli K1 PS. These findings strongly support the idea that the bacterium-specific epitope is the capsular PS or a structure invariantly associated with the capsular PS. An example of the invariably associated molecule is a lipid tail, which is used to anchor the capsular PS to the membrane and which is invariably found for the capsular PS of both E. coli and NMGB (4, 14). Also, PS-lipid binds to the ELISA plates much better than does pure PS (C. Frasch, personal communication). Thus, antibodies to PS-lipid would be efficiently detected by the ELISA even though PS-lipid is only a minor contaminant of the purified preparation of PS. The epitope is not likely a protein because the preparation of purified E. coli K1 PS has very little protein contamination (W. Vann, personal communication) and any such protein would have to be conserved among widely different bacteria (E. coli versus N. meningitidis). We are currently investigating the nature of the bacterium-specific epitope further using EV36 variants with mutations in the synthesis or transport of the capsular PS. EV36 is nonpathogenic, and its use greatly simplifies such epitope studies.

These bacterium-specific MAbs differed in their bactericidal potencies. HmenB2 is extremely potent, but some MAbs are not bactericidal at all. All our hybridomas produce IgM antibody, and a possible explanation is that some highly potent antibodies (e.g., HmenB2) may be hexameric IgM. Hexameric IgM has been found to fix complement better (28) and is more frequent among antibodies produced by peritoneal B cells (17), which are often associated with B1 B cells and anti-PS antibodies (16). A more likely possibility is that functional variability resides in the V region of antibodies. The V region differences may result in differences in the avidity or in the recognized epitopes. Two recent studies support the idea that the specificity of the epitope can be critically important for antibody function. Casadevall and his colleagues have found two types of IgM anticryptococcal antibodies (22): one type binds the fungus with a rim-type pattern and is protective, whereas the other type binds it with a puffy-type pattern and is not protective. Another example was shown by Fusco et al., who described an antibody that, in the presence of complement alone, can kill group B streptococcus, which is a gram-positive bacterium and is normally resistant to complement-mediated bacteriolysis (13). To examine the possibilities described above, additional work is in progress.

When the peptide mimotopes of NMGB PS were obtained with the bacterium-specific MAbs, the heterogeneity of their epitopes was readily shown by the mimotope sequences. The mimotopes obtained with one hybridoma readily demonstrate the sequence similarities, and the consensus sequences can be easily identified in some cases. For instance, SHxxxAF is the readily recognizable consensus sequence of HmenB3 mimotopes. Identical mimotope sequences are frequently found among the phage clones independently derived from one selecting MAb. For instance, the PhaB14L1 sequence is expressed in 5 of 22 independent clones selected with HmenB14. In contrast to the mimotopes produced with one selecting antibody, mimotope peptides obtained with different selecting antibodies were quite distinct. For example, amino acids A, F, and H are common for mimotopes only from HmenB3 but amino acids L, M, and Q are common for those from HmenB14. An exception was observed for the phage clones obtained with HmenB2 and HmenB3. The sequences of the mimotopes were often identical, and additional studies (data not shown) found them to have the identical V region sequences (and the two hybridomas must have been obtained from two different B cells sharing the clonal origin). This finding is different from those of Granoff et al., who obtained identical mimotopes using different selecting antibodies with different fine specificities of binding (15). While the reason for this difference is unclear, our study identifies a large number of independent mimotopes that may bind the antigen binding region of the NMGB PS-specific antibodies.

Our mimotope sequences are different from the sequences of the six mimotopes of NMGB PS reported by Moe et al. (23). For instance, R is prominent in their sequences but not in our sequences. R is common among many PS mimotopes and may mimic hydroxyl groups. Their sequences have cysteine, but ours do not. In contrast, amino acids P and S were common for the phage clones from all the hybridomas. An unusual sequence is one in PhaB13L11 which has five P's in succession and likely forms a helix like NMGB PS. We believe, however, that the helices are different since polyproline forms a collagen-like helix with a 9-Å pitch whereas NMGB PSA most likely forms a helix with a 5.5-Å pitch (6). Another interesting phage is PhaB13L4, which has a WSY sequence. The W/YXY sequence motif has been found in the peptide mimics of several PS (31) including group C PS of N. meningitidis, another PSA with an (2-9) linkage group (33). No other obvious similarities have been observed when our sequences were compared with mimotopes of other PS molecules.

The recent identification of new vaccine candidate molecules (26) has opened a new approach to making a vaccine against NMGB. However, these molecules have not been extensively studied so far, and the prospects for their becoming a successful vaccine are still unclear. In contrast, the capsular PS of NMGB has been extremely well studied, and the antibodies to NMGB PS have been shown to be effective against all strains of NMGB. There is also an increasing body of literature showing that mimotopes can serologically mimic carbohydrate epitopes and elicit antibodies to a variety of PS antigens including group C meningococcal PS (33), cryptococcal capsular PS (D. O. Beenhouwer, P. Valadon, R. May, S. L. Morrison, and M. D. Scharff, FASEB J. 14:A947, 2000), or carbohydrate epitopes of human immunodeficiency virus type 1 or respiratory syncytial virus (7) and other PS (2, 31). Thus, we believe that bacterium-specific anti-NMGB PS antibody can be induced with the mimotopes. We are currently investigating the immunogenicity of additional mimotopes and the interaction between the mimotope peptide and the antibody in detail.